Merge pull request #117 from Hestia-Homes/sap-dev

Sap dev

.dockerignore

@@ -0,0 +1,9 @@
+modules/ml-pipeline/src/pipeline/data/predictions
+modules/ml-pipeline/src/pipeline/data/fit_predictions
+modules/ml-pipeline/src/pipeline/data/prepared_data
+modules/ml-pipeline/src/pipeline/data/model/allmodels
+modules/ml-pipeline/src/pipeline/metrics
+modules/ml-pipeline/src/pipeline/__pycache__
+modules/ml-pipeline/src/pipeline/.dvc
+modules/ml-pipeline/src/pipeline/analysis
+modules/ml-pipeline/src/pipeline/metrics

.github/workflows/Deploy.yml

@@ -0,0 +1,127 @@
+name: Sap Change Model Deploy
+
+on:
+  push:
+    branches: [ sap-dev, sap-prod, heat-dev, heat-prod, carbon-dev, carbon-prod]
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v3
+
+      - name: Set up Python
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.10.12
+
+      - name: Install Serverless and plugins
+        run: |
+          npm install -g serverless@^3.38.0
+          npm install -g serverless-domain-manager@^7.3.8
+
+      - name: Install DVC
+        run: |
+          pip install --upgrade pip
+          pip install -r modules/ml-pipeline/src/pipeline/requirements/version_control/requirements.txt
+
+      # Set up all of the secrets required for the deployment
+      - name: set secret prefix which is used across multiple steps
+        id: secret_prefix
+        run: |
+            # Convert branch name to uppercase and replace hyphens with underscores
+            echo "::set-output name=secret_prefix::$(echo "${{ github.ref_name }}" | tr 'a-z-' 'A-Z_')"
+
+      - name: Set domain name
+        id: set_domain
+        run: echo "::set-output name=domain::${{ secrets[format('{0}_DOMAIN_NAME', steps.secret_prefix.outputs.secret_prefix)] }}"
+
+      - name: Set ECR credentials
+        id: set_ecr_credentials
+        run: |
+          # Fetch the secret using the secret prefix
+          echo "::set-output name=ecr_uri::${{ secrets[format('{0}_ECR_URI', steps.secret_prefix.outputs.secret_prefix)] }}"
+
+      - name: Set S3 buckets
+        id: set_s3_buckets
+        run: |
+          # Fetch the secret using the secret prefix
+          echo "::set-output name=data_bucket::${{ secrets[format('{0}_DATA_BUCKET', steps.secret_prefix.outputs.secret_prefix)] }}"
+          echo "::set-output name=predictions_bucket::${{ secrets[format('{0}_PREDICTIONS_BUCKET', steps.secret_prefix.outputs.secret_prefix)] }}"
+
+      - name: Set stack_name
+        id: set_stack_name
+        run: |
+          # Take branch prefix and add "model" for stack name
+          stack_name=$( echo ${{ github.ref_name }} | awk -F"-" '{print $1}' | sed 's/$/model/g')
+          if [ -z "${stack_name}" ]; then
+            echo "::set-output name=stack_name::"
+          else
+            echo "::set-output name=stack_name::${stack_name}"
+          fi
+
+      - name: Set runtime_environment
+        id: set_runtime_environment
+        run: |
+          # Extract the suffix after the hyphen from the branch name
+          runtime_environment=$(echo "${{ github.ref_name }}" | awk -F'-' '{print $NF}')
+          echo "::set-output name=runtime_environment::$runtime_environment"
+
+      - name: AWS credentials for dev
+        if: ${{ steps.set_runtime_environment.outputs.runtime_environment }} == 'dev'
+        uses: aws-actions/configure-aws-credentials@v1
+        with:
+          aws-access-key-id: ${{ secrets.DEV_AWS_ACCESS_KEY_ID }}
+          aws-secret-access-key: ${{ secrets.DEV_AWS_SECRET_ACCESS_KEY }}
+          aws-region: eu-west-2
+
+      - name: AWS credentials for prod
+        if: ${{ steps.set_runtime_environment.outputs.runtime_environment }} == 'prod'
+        uses: aws-actions/configure-aws-credentials@v1
+        with:
+          aws-access-key-id: ${{ secrets.PROD_AWS_ACCESS_KEY_ID }}
+          aws-secret-access-key: ${{ secrets.PROD_AWS_SECRET_ACCESS_KEY }}
+          aws-region: eu-west-2
+
+      - name: DVC Pull
+        run: |
+          cd modules/ml-pipeline/src/pipeline
+          dvc pull -r ${{ steps.set_runtime_environment.outputs.runtime_environment }}
+
+      - name: Setup Docker
+        uses: docker/setup-buildx-action@v1
+
+      - name: Login to ECR
+        run: |
+          aws ecr get-login-password --region eu-west-2 | docker login --username AWS --password-stdin ${{ steps.set_ecr_credentials.outputs.ecr_uri }}
+
+      # Building and pushing Docker image with caching
+      - name: Build and push Docker image
+        uses: docker/build-push-action@v3
+        with:
+          context: .
+          file: ./deployment/Dockerfile.prediction.lambda
+          push: true
+          tags: ${{ steps.set_ecr_credentials.outputs.ecr_uri }}:${{ github.sha }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
+          platforms: linux/amd64
+          provenance: false
+          build-args: |
+            RUNTIME_ENVIRONMENT=${{ steps.set_runtime_environment.outputs.runtime_environment }}
+
+      - name: Deploy to AWS Lambda via Serverless
+        env:
+          RUNTIME_ENVIRONMENT: ${{ steps.set_runtime_environment.outputs.runtime_environment }}
+          PREDICTIONS_BUCKET: ${{ steps.set_s3_buckets.outputs.predictions_bucket }}
+          DATA_BUCKET:  ${{ steps.set_s3_buckets.outputs.data_bucket }}
+          DOMAIN_NAME: ${{ steps.set_domain.outputs.domain }}
+          ECR_URI: ${{ steps.set_ecr_credentials.outputs.ecr_uri }}
+          GITHUB_SHA: ${{ github.sha }}
+          STACK_NAME: ${{ steps.set_stack_name.outputs.stack_name }}
+        run: |
+          # Deploy to AWS Lambda via Serverless
+          cd deployment
+          sls deploy --config serverless.yml --stage ${{ steps.set_runtime_environment.outputs.runtime_environment }} --verbose

.github/workflows/MLPipelinePostMerge.yml

@@ -10,7 +10,9 @@ on:
     types:
       - closed
     branches:
-      - "master"
+      - "sap-dev"
+      - "heat-dev"
+      - "carbon-dev"
 
 permissions: write-all
 
@@ -40,7 +42,14 @@ jobs:
         if [ -z "${latest_version}" ]; then
           increment_version="1.0.0"
         else
-          increment_version=$(echo ${latest_version} | awk -F'.' '{OFS="."; $1+=1; print}')
+          increment_version=$(echo ${latest_version} | awk 'BEGIN {
+              FS="\\."   # Set the field separator to a period
+              OFS="."    # Set the output field separator to a period
+          }
+          {
+              major = $1 + 1   # Increment the major version
+              print major, "0", "0"   # Print the new version
+          }')
         fi
 
         new_tag=${REGISTER_MODEL_NAME}@v${increment_version}
@@ -48,7 +57,7 @@ jobs:
         git tag -a ${new_tag} -m "Registering new Major Version"
         git push origin ${new_tag}
 
-        gto show > MODEL_REGISTRY.md
+        gto show --json > MODEL_REGISTRY.md
         git add .
         git commit -m "Update Registry"
         git push
@@ -78,7 +87,14 @@ jobs:
         if [ -z "${latest_version}" ]; then
           increment_version="0.1.0"
         else
-          increment_version=$(echo ${latest_version} | awk 'BEGIN{FS=OFS="."} {$2++; print}')
+          increment_version=$(echo ${latest_version} | awk 'BEGIN {
+              FS="\\."   # Set the field separator to a period
+              OFS="."    # Set the output field separator to a period
+          }
+          {
+              minor = $2 + 1   # Increment the minor version
+              print $1, minor, "0"   # Print the new version
+          }')
         fi
 
         new_tag=${REGISTER_MODEL_NAME}@v${increment_version}
@@ -86,7 +102,7 @@ jobs:
         git tag -a ${new_tag} -m "Registering new Minor Version"
         git push origin ${new_tag}
 
-        gto show > MODEL_REGISTRY.md
+        gto show --json > MODEL_REGISTRY.md
         git add .
         git commit -m "Update Registry"
         git push
@@ -116,7 +132,14 @@ jobs:
         if [ -z "${latest_version}" ]; then
           increment_version="0.0.1"
         else
-          increment_version=$(echo ${latest_version} | awk 'BEGIN{FS=OFS="."} {$3++; print}')
+          increment_version=$(echo ${latest_version} | awk 'BEGIN {
+              FS="\\."   # Set the field separator to a period
+              OFS="."    # Set the output field separator to a period
+          }
+          {
+              patch = $3 + 1   # Increment the patch version
+              print $1, $2, patch   # Print the new version
+          }')
         fi
 
         new_tag=${REGISTER_MODEL_NAME}@v${increment_version}
@@ -124,7 +147,7 @@ jobs:
         git tag -a ${new_tag} -m "Registering new Patch Version"
         git push origin ${new_tag}
 
-        gto show > MODEL_REGISTRY.md
+        gto show --json > MODEL_REGISTRY.md
         git add .
         git commit -m "Update Registry"
         git push
@@ -176,6 +199,8 @@ jobs:
         pip install -r modules/ml-pipeline/src/pipeline/requirements/version_control/requirements.txt
 
     - name: Register Model
+      env:
+        TARGET_BRANCH: ${{ github.base_ref }}
       run: |
 
         REGISTER_MODEL_NAME=$(echo ${{ github.event.pull_request.head.ref }} | awk -F"-" '{print $1}')
@@ -184,7 +209,7 @@ jobs:
         git config user.name "Github-Bot"
         git config user.email "Github-Bot@no-reply.com"
 
-        latest_dev_version=$(gto history ${REGISTER_MODEL_NAME} --asc --plain | awk '{print $NF}' | awk '/dev/')
+        latest_dev_version=$(gto history ${REGISTER_MODEL_NAME} --asc --plain | awk '{print $NF}' | awk '/dev/' | awk 'END {print}')
         if [ -z "${latest_dev_version}" ]; then
           increment_version="1"
         else
@@ -192,7 +217,7 @@ jobs:
         fi
 
         new_tag=${REGISTER_MODEL_NAME}#dev#${increment_version}
-        latest_version=$(gto show model@latest --ref | awk -F"@" '{print $2}')
+        latest_version=$(gto show ${REGISTER_MODEL_NAME}@latest --ref | awk -F"@" '{print $2}')
 
         echo ${new_tag}
 
@@ -203,11 +228,11 @@ jobs:
         git tag -a ${new_tag} -m "Assigning stage dev to artifact ${REGISTER_MODEL_NAME} version ${latest_version}"
         git push origin ${new_tag}
 
-        git checkout master
+        git checkout ${TARGET_BRANCH}
         git fetch --all
         git pull
 
-        gto show > MODEL_REGISTRY.md
+        gto show --json > MODEL_REGISTRY.md
         git add .
         git commit -m "Update Registry"
-        git push origin master
+        git push origin ${TARGET_BRANCH}

.github/workflows/MLPipelinePullRequest.yml

@@ -5,7 +5,7 @@ on:
   #   branches:
   #     - "model-**"
   pull_request:
-    branches: [ "master" ]
+    branches: ["sap-dev", "heat-dev", "carbon-dev"]
   label:
     types: ["created", "edited"]
 
@@ -89,13 +89,24 @@ jobs:
         AWS_ACCESS_KEY_ID: ${{ secrets.ROBOT_AWS_ACCESS_KEY_ID }}
         AWS_SECRET_ACCESS_KEY: ${{ secrets.ROBOT_AWS_SECRET_ACCESS_KEY }}
         REPO_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        TARGET_BRANCH: ${{ github.base_ref }}
       run: |
         cd modules/ml-pipeline/src/pipeline
         echo "## Model metrics" > report.md
 
         # Compare metrics to master
-        git fetch --depth=1 origin master:master
-        dvc metrics diff --md --all master >> report.md
+        git fetch --depth=1 origin ${TARGET_BRANCH}:${TARGET_BRANCH}
+        dvc metrics diff --md --all ${TARGET_BRANCH} >> report.md
+
+        echo "## Scenario comparison" >> report.md
+
+        cat metrics/scenario_table.md >> report.md
+
+        echo "" >> report.md
+
+        echo "## Scenario metrics" >> report.md
+
+        cat metrics/scenario_metrics.md >> report.md
 
         cml comment create report.md
 

MODEL_REGISTRY.md

@@ -1,5 +1,34 @@
-╒════════╤══════════╤═════════╕
-│ name   │ latest   │ #dev    │
-╞════════╪══════════╪═════════╡
-│ model  │ v10.10.  │ v10.10. │
-╘════════╧══════════╧═════════╛
+{
+    "model": {
+        "version": "v12.10.12",
+        "stage": {
+            "dev": "v11.10.12"
+        },
+        "registered": true,
+        "active": true
+    },
+    "sap": {
+        "version": "v0.14.0",
+        "stage": {
+            "dev": "v0.14.0"
+        },
+        "registered": true,
+        "active": true
+    },
+    "heat": {
+        "version": "v0.5.0",
+        "stage": {
+            "dev": "v0.5.0"
+        },
+        "registered": true,
+        "active": true
+    },
+    "carbon": {
+        "version": "v0.5.0",
+        "stage": {
+            "dev": "v0.5.0"
+        },
+        "registered": true,
+        "active": true
+    }
+}

README.md

@@ -3,21 +3,83 @@
 Creating a ML-toolkit that can be reused:
 
 - ML pipeline:
-	- A generic pipeline that has data version control, experiment 
+	- A generic pipeline that has data version control, experiment
 tracking and a model registry
 
 - ML monitoring:
 	- A bolt-on service that can implement model monitoring
 
 There are multiple protected branches which adapt the generic pipeline to produce different models:
-- sap_change-**
-- heat_change-**
-- carbon_change-**
+- sap-{dev/staging/prod}-**
+- heat-{dev/staging/prod}-**
+- carbon-{dev/staging/prod}-**
 
 These branches will differ by the configuration files that define the data used and the outputs of the ML-pipeline
 - There can be different additional logic for each branch but the pipeline will be the same.
 
-# Deployment 
+# Deployment
 
-TBD
+Scripts associated to deployment can be found in the deployment/ folder.
 
+Deployment is automated via Github Actions, where a deployment is triggered by a push to one of the
+protected branch, with one of dev or prod as the suffix, describing the target environment.
+
+The github actions file will build and push a docker image to ECR and then deploy a lambda
+which produces predictions for the relevant model.
+
+In order for this to be set up, some key environment variables needs to be inserted into Github
+secrets. Each different model and protected branch has its own set of secrets which allows for flexibility
+between different pipelines.
+
+For example, for the branch sap-dev, the prefix=SAP_DEV, and the following secrets are:
+
+- {prefix}_ECR_URI, which is the URI of the ECR repository to push to. For example, for the
+   sap change model this is the lambda-sap-prediction-dev repository.
+- {prefix}_DOMAIN_NAME, is the custom domain name. This is likely going to be the same across the different
+   models, but is still included in the secrets for flexibility.
+- {prefix}_DATA_BUCKET, is the name of the s3 data bucket where data to be scored by the model is stored
+- {prefix}_MODEL_BUCKET, is the name of the s3 bucket where the model is stored
+- {prefix}_PREDICTIONS_BUCKET, is the name of the s3 bucket where the predictions are stored
+
+
+# Building and Testing the Prediction Lambda Function Locally
+TODO: Generalise these instructions for the various different pipelines
+
+This guide outlines the steps to build and test the Lambda function locally using Docker. These instructions assume you're working with a machine that has Docker installed.
+
+### Prerequisites
+Docker: Make sure Docker is installed and running on your machine.
+AWS Credentials: Ensure you have AWS credentials set up on your local machine, typically stored
+in ~/.aws/credentials.
+Root Directory: All commands should be run from the root directory of the repository.
+Step-by-Step Guide
+1. Building the Docker Image
+First, navigate to the root directory of the repository. Open a terminal and execute the following
+2. command to build the Docker image:
+
+```bash
+docker build -t sap -f deployment/Dockerfile.prediction.lambda .
+```
+
+This will build a Docker image tagged as sap_change using the Dockerfile.prediction.lambda located
+in the deployment directory.
+
+2. Running the Docker Image
+Once the image is built, you can run it using the following command:
+
+```bash
+docker run -p 9000:8080 -v ~/.aws/credentials:/root/.aws/credentials:ro -e RUNTIME_ENVIRONMENT=dev -e PREDICTIONS_BUCKET=retrofit-sap-predictions-dev sap
+```
+This command does the following:
+
+Maps port 9000 on your local machine to port 8080 on the Docker container.
+Mounts your AWS credentials into the Docker container in read-only mode.
+Sets the RUNTIME_ENVIRONMENT variable to dev.
+3. Testing the Lambda Function
+To test the Lambda function, use the following curl command:
+
+```json
+curl -XPOST "http://localhost:9000/2015-03-31/functions/function/invocations" -d '{"body": "{\"file_location\": \"s3://retrofit-data-dev/sap_change_model/one_sample_test_dataset.parquet\", \"property_id\": 1, \"portfolio_id\": 4, \"created_at\": \"now\"}"}'
+```
+
+This will send a POST request to the running Lambda function and pass in the required data as JSON.

deployment/.dockerignore

@@ -0,0 +1,9 @@
+modules/ml-pipeline/src/pipeline/data/predictions
+modules/ml-pipeline/src/pipeline/data/fit_predictions
+modules/ml-pipeline/src/pipeline/data/prepared_data
+modules/ml-pipeline/src/pipeline/data/model/allmodels
+modules/ml-pipeline/src/pipeline/metrics
+modules/ml-pipeline/src/__pycache__
+modules/ml-pipeline/src/.dvc
+modules/ml-pipeline/src/analysis
+modules/ml-pipeline/src/metrics

deployment/Dockerfile.prediction.lambda

@@ -0,0 +1,25 @@
+FROM public.ecr.aws/lambda/python:3.10
+
+# Set the working directory
+WORKDIR ${LAMBDA_TASK_ROOT}
+ENV PYTHONPATH "${PYTHONPATH}:${LAMBDA_TASK_ROOT}"
+
+# Environment variables
+ARG RUNTIME_ENVIRONMENT
+ENV RUNTIME_ENVIRONMENT=${RUNTIME_ENVIRONMENT}
+
+# Install necessary build tools - required to test locally
+RUN yum install -y gcc python3-devel gcc-c++
+
+# Install python packages
+COPY modules/ml-pipeline/src/pipeline/requirements/predictions/requirements.txt ./requirements.txt
+RUN pip install --no-cache-dir -r ./requirements.txt
+
+# Copy the project code
+COPY modules/ml-pipeline/src/pipeline ./pipeline
+# Copy the handler
+COPY deployment/handlers/prediction_app.py ./pipeline/prediction_app.py
+WORKDIR ${LAMBDA_TASK_ROOT}/pipeline
+
+
+CMD [ "prediction_app.handler" ]

deployment/handlers/prediction_app.py

@@ -0,0 +1,123 @@
+"""
+This script is the handler for the lambda prediction function, responsible
+for producting predictions for a model
+"""
+
+import boto3
+from botocore.exceptions import NoCredentialsError
+import json
+from io import StringIO
+import os
+import logging
+from generate_predictions import generate_predictions
+from core.MLModels import model_factory
+from config import settings
+from core.DataClient import dataclient_factory
+
+logger = logging.getLogger()
+logger.setLevel(logging.INFO)
+
+PREDICTIONS_BUCKET = os.getenv("PREDICTIONS_BUCKET", None)
+
+
+def upload_dataframe_to_s3(df, bucket, s3_file_name):
+    """
+    Upload a pandas DataFrame to an S3 bucket as CSV
+
+    :param df: DataFrame to upload
+    :param bucket: Bucket to upload to
+    :param s3_file_name: S3 object name
+    :return: True if file was uploaded, else False
+    """
+
+    # Initialize the S3 client
+    s3 = boto3.client("s3")
+    csv_buffer = StringIO()
+
+    # Write the DataFrame to the buffer as CSV
+    df.to_csv(csv_buffer, index=False)
+
+    try:
+        # Upload the CSV from the buffer to S3
+        s3.put_object(Bucket=bucket, Key=s3_file_name, Body=csv_buffer.getvalue())
+        print(f"Successfully uploaded DataFrame to {bucket}/{s3_file_name}")
+        return True
+    except NoCredentialsError:
+        print("Credentials not available")
+        return False
+
+
+def handler(event, context):
+    """
+    Take in event and trigger the prediction pipeline
+    """
+
+    logger.info("received event: " + str(event))
+
+    try:
+        body = (
+            json.loads(event["body"])
+            if not isinstance(event["body"], dict)
+            else event["body"]
+        )
+
+        property_id = body["property_id"]
+        portfolio_id = body["portfolio_id"]
+        created_at = body["created_at"]
+
+        # TODO: Implement the loading of the model and prediction
+
+        storage_filepath = f"s3://{PREDICTIONS_BUCKET}/{portfolio_id}/{property_id}/{created_at}.parquet"
+
+        logger.info(f"--- Initiate MLModel ---")
+
+        build_model_params = settings.build_model
+        client_params = settings.client
+        feature_process_params = settings.feature_processor
+        generate_predictions_params = settings.generate_predictions
+
+        model = model_factory(build_model_params["model_type"])
+
+        logger.info(f"--- Initiate Input DataClient ---")
+        input_dataclient = dataclient_factory(
+            dataclient_type="aws-s3",
+            dataclient_config=client_params["aws-s3"],
+        )
+
+        logger.info(f"--- Initiate Output DataClient ---")
+        output_dataclient = dataclient_factory(
+            dataclient_type="aws-s3",
+            dataclient_config=client_params["aws-s3"],
+        )
+
+        generate_predictions(
+            input_dataclient=input_dataclient,
+            output_dataclient=output_dataclient,
+            model=model,
+            target=feature_process_params["feature_processor_config"]["target"],
+            model_filepath=build_model_params["model_save_filepath"],
+            test_data_filepath=body["file_location"],
+            predictions_output_filepath=storage_filepath,
+            predictions_column_name=generate_predictions_params[
+                "predictions_column_name"
+            ],
+            identifier_column=generate_predictions_params["identifier_column"],
+        )
+
+        return {
+            "statusCode": 200,
+            "body": json.dumps(
+                {
+                    "message": "Successfully processed input",
+                    "storage_filepath": storage_filepath,
+                }
+            ),
+        }
+
+    except (Exception, KeyError, ValueError) as e:
+        logger.info("Prediction failed")
+        logger.info(e)
+        return {
+            "statusCode": 500,
+            "body": json.dumps({"message": "Prediction failed", "error": str(e)}),
+        }

deployment/serverless.yml

@@ -0,0 +1,53 @@
+service: ${env:STACK_NAME}
+
+provider:
+  name: aws
+  region: eu-west-2
+  architecture: x86_64
+  environment:
+    RUNTIME_ENVIRONMENT: ${env:RUNTIME_ENVIRONMENT}
+    PREDICTIONS_BUCKET: ${env:PREDICTIONS_BUCKET}
+    DATA_BUCKET: ${env:DATA_BUCKET}
+    DOMAIN_NAME: ${env:DOMAIN_NAME}
+    ECR_URI: ${env:ECR_URI}
+    GITHUB_SHA: ${env:GITHUB_SHA}
+  iam:
+    role:
+      name: ${env:STACK_NAME}_s3_access
+      statements:
+        # Allow reading from the DATA_BUCKET
+        - Effect: Allow
+          Action:
+            - s3:*
+          Resource:
+            - arn:aws:s3:::${env:DATA_BUCKET}
+            - arn:aws:s3:::${env:DATA_BUCKET}/*
+        # Allow reading and writing to PREDICTIONS_BUCKET
+        - Effect: Allow
+          Action:
+            - s3:*
+          Resource:
+            - arn:aws:s3:::${env:PREDICTIONS_BUCKET}
+            - arn:aws:s3:::${env:PREDICTIONS_BUCKET}/*
+
+
+
+plugins:
+  - serverless-domain-manager
+
+custom:
+  customDomain:
+    domainName: api.${self:provider.environment.DOMAIN_NAME}
+    basePath: ${env:STACK_NAME}
+    createRoute53Record: true
+    certificateArn: ${ssm:/ssl_certificate_arn}
+
+functions:
+  sap_prediction_lambda:
+    image:
+      uri: ${env:ECR_URI}:${env:GITHUB_SHA}
+    events:
+      - http:
+          path: /predict
+          method: POST
+    timeout: 120 # Set max run time to 2 minutes - we shouldn't need this much time so this can be reviewed

modules/ml-pipeline/.dvc/.gitignore

@@ -1,3 +0,0 @@
-/config.local
-/tmp
-/cache

modules/ml-pipeline/.dvc/config

@@ -1,2 +0,0 @@
-['remote "myremote"']
-    url = /tmp/dvcstore

modules/ml-pipeline/.dvcignore

@@ -1,3 +0,0 @@
-# Add patterns of files dvc should ignore, which could improve
-# the performance. Learn more at
-# https://dvc.org/doc/user-guide/dvcignore

modules/ml-pipeline/.gitignore

@@ -3,3 +3,4 @@
 __pycache__/
 .DS_Store
 .vscode/
+data/

modules/ml-pipeline/.gto

@@ -1,2 +0,0 @@
-# .gto config file
-stages: [dev, stage, prod] # list of allowed Stages

modules/ml-pipeline/Makefile

@@ -9,16 +9,16 @@ init: dev-conda
 .PHONY: dev-conda
 dev-conda:
 	# conda deactivate || echo "Not in conda environment"
-	# conda remove --name $CONDA_ENV --all -y || echo "No environment created previously"
-	conda create --name $CONDA_ENV python=$(PYTHON_VERSION) -y
+	# conda remove --name ${CONDA_ENV} --all -y || echo "No environment created previously"
+	conda create --name ${CONDA_ENV} python=$(PYTHON_VERSION) -y
 	conda init bash
-	conda run -vvvv -n $CONDA_ENV pip install --upgrade pip
-	conda run -vvvv -n $CONDA_ENV pip install -r src/pipeline/requirements/training/requirements-dev.txt
-	conda run -vvvv -n $CONDA_ENV pip install -r src/pipeline/requirements/version_control/requirements.txt
-	conda run -vvvv -n $CONDA_ENV pre-commit install
-	conda run -vvvv -n $CONDA_ENV pip install ipykernel
+	conda run -v -n ${CONDA_ENV} pip install --upgrade pip
+	conda run -v -n ${CONDA_ENV} pip install -r src/pipeline/requirements/training/requirements-dev.txt
+	conda run -v -n ${CONDA_ENV} pip install -r src/pipeline/requirements/version_control/requirements.txt
+	conda run -v -n ${CONDA_ENV} pre-commit install
+	conda run -v -n ${CONDA_ENV} pip install ipykernel
 	echo "TO ACTIVATE ENVIRONMENT, USE THE FOLLOWING COMMAND"
-	echo "conda activate $CONDA_ENV"
+	echo "conda activate ${CONDA_ENV}"
 
 
 .PHONY: dev-pyenv

modules/ml-pipeline/src/.dockerignore

@@ -0,0 +1,8 @@
+pipeline/data/predictions
+pipeline/data/fit_predictions
+pipeline/data/prepared_data/train.parquet
+pipeline/data/fit_predictions
+pipeline/data/model/allmodels
+pipeline/metrics
+pipeline/.dvc
+pipeline/analysis

modules/ml-pipeline/src/Prediction.Dockerfile

@@ -1,7 +1,7 @@
 # Dockerfile that can be used to test loading a model to generate a prediction (part of CI/CD flow)
 FROM python:3.10.12-slim
 
-RUN apt-get update && apt-get install -y libgomp1
+RUN apt-get update && apt-get install -y libgomp1 gcc python3-dev
 
 COPY pipeline/requirements/predictions/requirements.txt requirements.txt
 

modules/ml-pipeline/src/README.md

@@ -1,3 +1,3 @@
 # The generic reproducible ML-pipeline
 
-Pipeline required to build a model to produce an output
+Pipeline required to build a model to produce an output, that gets hashed via DVC

modules/ml-pipeline/src/pipeline/0_startup_cleanup.py

@@ -16,13 +16,9 @@ def run_cleanup(artefacts_directory: str, metrics_directory: str) -> None:
     Remove the directory where artefacts are stored
     """
 
-    logger.info("---------------------")
     logger.info(f"--- Run Clean up ---")
-    logger.info("---------------------")
 
-    logger.info("-------------------------")
     logger.info(f"--- Delete artefacts ---")
-    logger.info("-------------------------")
 
     artefact_directory_path = Path(artefacts_directory)
 
@@ -31,9 +27,7 @@ def run_cleanup(artefacts_directory: str, metrics_directory: str) -> None:
         logger.info(f"Removing the directory: {artefacts_directory}")
         shutil.rmtree(artefact_directory_path)
 
-    logger.info("-----------------------")
     logger.info(f"--- Delete metrics ---")
-    logger.info("-----------------------")
 
     metrics_directory_path = Path(metrics_directory)
 
@@ -45,15 +39,11 @@ def run_cleanup(artefacts_directory: str, metrics_directory: str) -> None:
 
 if __name__ == "__main__":
 
-    logger.info("----------------------------")
     logger.info(f"--- {__file__} - Start! ---")
-    logger.info("----------------------------")
 
     run_cleanup(
         artefacts_directory=startup_cleanup_params["artefacts"],
         metrics_directory=startup_cleanup_params["metrics"],
     )
 
-    logger.info("-------------------------------")
     logger.info(f"--- {__file__} - Complete! ---")
-    logger.info("-------------------------------")

modules/ml-pipeline/src/pipeline/1_prepare_data.py

@@ -17,9 +17,7 @@ from core.DataClient import dataclient_factory
 from core.FeatureProcessor import feature_processor_factory
 from config import settings
 
-logger.info("----------------------------")
 logger.info(f"--- Initiate Parameters ---")
-logger.info("----------------------------")
 
 RUNTIME_ENVIRONMENT = os.environ.get("RUNTIME_ENVIRONMENT", "local")
 
@@ -33,9 +31,7 @@ output_train_filepath = prepare_data_params["output_train_filepath"]
 output_test_filepath = prepare_data_params["output_test_filepath"]
 feature_processor_config = feature_process_params["feature_processor_config"]
 
-logger.info("----------------------------")
 logger.info(f"--- Initiate DataClient ---")
-logger.info("----------------------------")
 
 input_dataclient_type = prepare_data_params["input_dataclient_type"]
 output_dataclient_type = prepare_data_params["output_dataclient_type"]
@@ -49,9 +45,7 @@ output_dataclient = dataclient_factory(
     dataclient_config=client_params[output_dataclient_type],
 )
 
-logger.info("----------------------------------")
 logger.info(f"--- Initiate FeatureProcessor ---")
-logger.info("----------------------------------")
 
 feature_processor = feature_processor_factory(
     feature_process_params["feature_processor_type"]
@@ -76,15 +70,11 @@ def prepare_data(
     :param pipeline_mode: bool, Default False, this caches out the file for experimentation, objects returned in pipeline mode
     """
 
-    logger.info("--------------------")
     logger.info("--- Loading data ---")
-    logger.info("--------------------")
 
     data = input_dataclient.load_data(location=data_filepath, load_config={})
 
-    logger.info("--------------------------")
     logger.info("--- Feature Processing ---")
-    logger.info("--------------------------")
 
     data = feature_processor.feature_process(
         data,
@@ -93,13 +83,12 @@ def prepare_data(
         new_feature_funcs=new_feature_funcs,
     )
 
-    logger.info("----------------------")
     logger.info("--- Splitting data ---")
-    logger.info("----------------------")
 
     if train_proportion == 1:
         train = data
-        test = None
+        # Sample 10% of the data for testing
+        test = data.sample(round(len(data) * 0.1))
     else:
         train, test = train_test_split(
             data, train_size=train_proportion, test_size=(1 - train_proportion)
@@ -108,9 +97,7 @@ def prepare_data(
 
     train = train.reset_index(drop=True)
 
-    logger.info("-----------------------")
     logger.info("--- Outputting data ---")
-    logger.info("-----------------------")
 
     output_dataclient.save_data(
         obj=train, location=output_train_filepath, save_config=None
@@ -126,13 +113,9 @@ def prepare_data(
 
 if __name__ == "__main__":
 
-    logger.info("----------------------------")
     logger.info(f"--- {__file__} - Start! ---")
-    logger.info("----------------------------")
 
-    logger.info("---------------------------")
     logger.info(f"--- Prepare Data Stage ---")
-    logger.info("---------------------------")
 
     prepare_data(
         input_dataclient=input_dataclient,
@@ -147,6 +130,4 @@ if __name__ == "__main__":
         new_feature_funcs=new_feature_funcs,
     )
 
-    logger.info("-------------------------------")
     logger.info(f"--- {__file__} - Complete! ---")
-    logger.info("-------------------------------")

modules/ml-pipeline/src/pipeline/2_build_model.py

@@ -6,7 +6,7 @@ Once we have the features, we build a model
 import os
 import yaml
 import pandas as pd
-from typing import Union
+from typing import Union, List
 from pathlib import Path
 from core.Logger import logger
 from core.interface.InterfaceMetrics import MLMetrics
@@ -18,9 +18,7 @@ from core.MLMetrics import metrics_factory
 from configs.post_prediction_logic import post_prediction_logic
 from config import settings
 
-logger.info("----------------------------")
 logger.info(f"--- Initiate Parameters ---")
-logger.info("----------------------------")
 
 RUNTIME_ENVIRONMENT = os.environ.get("RUNTIME_ENVIRONMENT", "local")
 
@@ -28,31 +26,31 @@ prepare_data_params = settings.prepare_data
 build_model_params = settings.build_model
 feature_process_params = settings.feature_processor
 generate_metrics_params = settings.generate_metrics
+generate_predictions_params = settings.generate_predictions
 
 model_type = build_model_params["model_type"]
 target = feature_process_params["feature_processor_config"]["target"]
+fit_predictions_filepath = build_model_params["fit_predictions_filepath"]
+predictions_column_name = generate_predictions_params["predictions_column_name"]
+identifier_columns = feature_process_params["feature_processor_config"][
+    "identifier_columns"
+]
 model_save_location = build_model_params["model_save_filepath"]
 model_hyperparameters = build_model_params[model_type]
 train_filepath = prepare_data_params["output_train_filepath"]
 test_filepath = prepare_data_params["output_test_filepath"]
 fit_metrics_filepath = build_model_params["fit_metrics_filepath"]
 
-logger.info("----------------------------")
 logger.info(f"--- Initiate DataClient ---")
-logger.info("----------------------------")
 
 # Output of previous prepare data step, will be where the data is
 dataclient = dataclient_factory(prepare_data_params["output_dataclient_type"])
 
-logger.info("-------------------------")
 logger.info(f"--- Initiate MLModel ---")
-logger.info("-------------------------")
 
 model = model_factory(model_type)
 
-logger.info("-------------------------")
 logger.info(f"--- Initiate Metrics ---")
-logger.info("-------------------------")
 
 metrics = metrics_factory(generate_metrics_params["metrics_type"])
 
@@ -62,8 +60,11 @@ def build_model(
     model: MLModel,
     metrics: MLMetrics,
     target: str,
+    identifier_columns: List[str],
     model_save_location: str,
     model_hyperparameters: dict,
+    fit_predictions_filepath: str,
+    predictions_column_name: str,
     fit_metrics_filepath: str,
     train_filepath: Union[str, None] = None,
     test_filepath: Union[str, None] = None,
@@ -71,9 +72,7 @@ def build_model(
     test_data: Union[pd.DataFrame, None] = None,
     pipeline_mode: bool = False,
 ):
-    logger.info("--------------------------------------")
     logger.info("--- Loading Data for build process ---")
-    logger.info("--------------------------------------")
 
     if train_data is None:
         if train_filepath is None:
@@ -85,42 +84,41 @@ def build_model(
             raise ValueError(f"Need {test_filepath} if no data supplied")
         test_data = dataclient.load_data(location=test_filepath, load_config=None)
 
-    logger.info("----------------------")
     logger.info("--- Training model ---")
-    logger.info("----------------------")
 
     model.train_model(
-        data=train_data, target=target, model_hyperparameters=model_hyperparameters
+        data=train_data.drop(columns=identifier_columns),
+        target=target,
+        model_hyperparameters=model_hyperparameters,
     )
 
-    logger.info("----------------------------------")
     logger.info("--- Generating fit predictions ---")
-    logger.info("----------------------------------")
-
-    prediction_data = train_data.drop(columns=target)
 
     fit_predictions = model.predict(
-        data=prediction_data, post_prediction_logic=post_prediction_logic
+        data=train_data, post_prediction_logic=post_prediction_logic
+    )
+
+    logger.info("--- Saving fit predictions ---")
+
+    predictions_df = pd.DataFrame(fit_predictions)
+    predictions_df.columns = [predictions_column_name]
+
+    dataclient.save_data(
+        obj=predictions_df, location=fit_predictions_filepath, save_config=None
     )
 
-    logger.info("------------------------------")
     logger.info("--- Generating fit metrics ---")
-    logger.info("------------------------------")
 
     metrics_output = metrics.generate_metrics(
         target=train_data[target],
         predictions=pd.Series(fit_predictions),
     )
 
-    logger.info("--------------------")
     logger.info("--- Saving model ---")
-    logger.info("--------------------")
 
     model.save_model(path=Path(model_save_location))
 
-    logger.info("--------------------------")
     logger.info("--- Saving fit metrics ---")
-    logger.info("--------------------------")
 
     dataclient.save_data(
         obj=metrics_output, location=fit_metrics_filepath, save_config=None
@@ -129,26 +127,23 @@ def build_model(
 
 if __name__ == "__main__":
 
-    logger.info("----------------------------")
     logger.info(f"--- {__file__} - Start! ---")
-    logger.info("----------------------------")
 
-    logger.info("--------------------------")
     logger.info(f"--- Build Model Stage ---")
-    logger.info("--------------------------")
 
     build_model(
         dataclient=dataclient,
         model=model,
         metrics=metrics,
         target=target,
+        identifier_columns=identifier_columns,
         model_save_location=model_save_location,
         model_hyperparameters=model_hyperparameters,
         train_filepath=train_filepath,
         test_filepath=test_filepath,
         fit_metrics_filepath=fit_metrics_filepath,
+        fit_predictions_filepath=fit_predictions_filepath,
+        predictions_column_name=predictions_column_name,
     )
 
-    logger.info("-------------------------------")
     logger.info(f"--- {__file__} - Complete! ---")
-    logger.info("-------------------------------")

modules/ml-pipeline/src/pipeline/3_generate_predictions.py

@@ -4,20 +4,13 @@ After the model is built, we can evaluate its performance
 """
 
 import os
-import yaml
-import pandas as pd
-from pathlib import Path
-from core.interface.InterfaceModels import MLModel
-from core.interface.InterfaceDataClient import DataClient
 from core.DataClient import dataclient_factory
 from core.MLModels import model_factory
 from core.Logger import logger
-from configs.post_prediction_logic import post_prediction_logic
 from config import settings
+from generate_predictions import generate_predictions
 
-logger.info("----------------------------")
 logger.info(f"--- Initiate Parameters ---")
-logger.info("----------------------------")
 
 RUNTIME_ENVIRONMENT = os.environ.get("RUNTIME_ENVIRONMENT", "local")
 
@@ -38,15 +31,11 @@ model_filepath = build_model_params["model_save_filepath"]
 predictions_output_filepath = generate_predictions_params["predictions_output_filepath"]
 predictions_column_name = generate_predictions_params["predictions_column_name"]
 
-logger.info("-------------------------")
 logger.info(f"--- Initiate MLModel ---")
-logger.info("-------------------------")
 
 model = model_factory(build_model_params["model_type"])
 
-logger.info("----------------------------")
 logger.info(f"--- Initiate DataClient ---")
-logger.info("----------------------------")
 
 # We may have different locations of loading hence why we use one specified in generate_predictions.yaml
 # I.e. for metric runs, this will be a local data client
@@ -62,67 +51,11 @@ output_dataclient = dataclient_factory(
 )
 
 
-def generate_predictions(
-    input_dataclient: DataClient,
-    output_dataclient: DataClient,
-    model: MLModel,
-    target: str,
-    model_filepath: str,
-    test_data_filepath: str,
-    predictions_output_filepath: str,
-    predictions_column_name: str,
-):
-    """
-    For a given model, we generate prediction and evaluate this against the true target
-    """
-
-    logger.info("-------------------------")
-    logger.info("--- Loading test data ---")
-    logger.info("-------------------------")
-
-    test_data = input_dataclient.load_data(
-        location=test_data_filepath, load_config=None
-    )
-
-    logger.info("---------------------")
-    logger.info("--- Loading model ---")
-    logger.info("---------------------")
-
-    model.load_model(model_filepath)
-
-    logger.info("------------------------------")
-    logger.info("--- Generating predictions ---")
-    logger.info("------------------------------")
-
-    prediction_data = (
-        test_data.drop(columns=target) if target in test_data.columns else test_data
-    )
-
-    predictions = model.predict(
-        data=prediction_data, post_prediction_logic=post_prediction_logic
-    )
-
-    logger.info("--------------------------")
-    logger.info("--- Saving predictions ---")
-    logger.info("--------------------------")
-
-    predictions_df = pd.DataFrame(predictions)
-    predictions_df.columns = [predictions_column_name]
-
-    output_dataclient.save_data(
-        obj=predictions_df, location=predictions_output_filepath, save_config=None
-    )
-
-
 if __name__ == "__main__":
 
-    logger.info("----------------------------")
     logger.info(f"--- {__file__} - Start! ---")
-    logger.info("----------------------------")
 
-    logger.info("----------------------------------")
     logger.info(f"--- Generate Predictions Stage---")
-    logger.info("----------------------------------")
 
     generate_predictions(
         input_dataclient=input_dataclient,
@@ -135,6 +68,4 @@ if __name__ == "__main__":
         predictions_column_name=predictions_column_name,
     )
 
-    logger.info("-------------------------------")
     logger.info(f"--- {__file__} - Complete! ---")
-    logger.info("-------------------------------")

modules/ml-pipeline/src/pipeline/4_generate_metrics.py

@@ -16,9 +16,7 @@ from core.MLMetrics import metrics_factory
 from core.Logger import logger
 from config import settings
 
-logger.info("----------------------------")
 logger.info(f"--- Initiate Parameters ---")
-logger.info("----------------------------")
 
 RUNTIME_ENVIRONMENT = os.environ.get("RUNTIME_ENVIRONMENT", "local")
 
@@ -35,16 +33,11 @@ predictions_output_filepath = generate_predictions_params["predictions_output_fi
 predictions_column_name = generate_predictions_params["predictions_column_name"]
 metrics_output_filepath = generate_metrics_params["metrics_output_filepath"]
 
-
-logger.info("-------------------------")
 logger.info(f"--- Initiate MLModel ---")
-logger.info("-------------------------")
 
 model = model_factory(build_model_params["model_type"])
 
-logger.info("----------------------------")
 logger.info(f"--- Initiate DataClient ---")
-logger.info("----------------------------")
 
 # Use data client for input and output, as we use dvc to cache later to the cloud
 dataclient_type = generate_metrics_params["dataclient_type"]
@@ -53,9 +46,7 @@ dataclient = dataclient_factory(
     dataclient_config=client_params[dataclient_type],
 )
 
-logger.info("---------------------------")
 logger.info(f"--- Initiate MLMetrics ---")
-logger.info("---------------------------")
 
 metrics = metrics_factory(generate_metrics_params["metrics_type"])
 
@@ -75,34 +66,26 @@ def generate_metrics(
     For a given model, we generate prediction and evaluate this against the true target
     """
 
-    logger.info("-------------------------")
     logger.info("--- Loading test data ---")
-    logger.info("-------------------------")
 
     test_data = input_dataclient.load_data(
         location=test_data_filepath, load_config=None
     )
 
-    logger.info("---------------------------")
     logger.info("--- Loading predictions ---")
-    logger.info("---------------------------")
 
     predictions = input_dataclient.load_data(
         location=predictions_output_filepath, load_config=None
     )
 
-    logger.info("--------------------------")
     logger.info("--- Generating metrics ---")
-    logger.info("--------------------------")
 
     metrics_output = metrics.generate_metrics(
         target=test_data[target],
         predictions=pd.Series(predictions[predictions_column_name]),
     )
 
-    logger.info("----------------------")
     logger.info("--- Saving metrics ---")
-    logger.info("----------------------")
 
     output_dataclient.save_data(
         obj=metrics_output, location=metrics_output_filepath, save_config=None
@@ -111,13 +94,9 @@ def generate_metrics(
 
 if __name__ == "__main__":
 
-    logger.info("----------------------------")
     logger.info(f"--- {__file__} - Start! ---")
-    logger.info("----------------------------")
 
-    logger.info("------------------------------")
     logger.info(f"--- Generate Metrics Stage---")
-    logger.info("------------------------------")
 
     generate_metrics(
         input_dataclient=dataclient,
@@ -131,6 +110,4 @@ if __name__ == "__main__":
         metrics_output_filepath=metrics_output_filepath,
     )
 
-    logger.info("-------------------------------")
     logger.info(f"--- {__file__} - Complete! ---")
-    logger.info("-------------------------------")

modules/ml-pipeline/src/pipeline/5_generate_scenarios.py

@@ -0,0 +1,162 @@
+"""
+Fourth part of the pipeline:
+After the model is built and metrics are generated,
+we want to test this model against known scenarios
+"""
+
+import os
+import pandas as pd
+from core.interface.InterfaceModels import MLModel
+from core.interface.InterfaceDataClient import DataClient
+from core.interface.InterfaceMetrics import MLMetrics
+from configs.post_prediction_logic import post_prediction_logic
+from core.DataClient import dataclient_factory
+from core.MLModels import model_factory
+from core.MLMetrics import metrics_factory
+from core.Logger import logger
+from config import settings
+
+logger.info(f"--- Initiate Parameters ---")
+
+RUNTIME_ENVIRONMENT = os.environ.get("RUNTIME_ENVIRONMENT", "local")
+
+client_params = settings.client
+prepare_data_params = settings.prepare_data
+build_model_params = settings.build_model
+generate_predictions_params = settings.generate_predictions
+generate_metrics_params = settings.generate_metrics
+feature_process_params = settings.feature_processor
+scenarios_params = settings.scenarios
+
+model_filepath = build_model_params["model_save_filepath"]
+target = feature_process_params["feature_processor_config"]["target"]
+scenario_data_filepaths = scenarios_params["scenario_data_filepaths"]
+predictions_column_name = generate_predictions_params["predictions_column_name"]
+comparison_output_filepath = scenarios_params["comparison_output_filepath"]
+metrics_output_filepath = scenarios_params["metrics_output_filepath"]
+
+logger.info(f"--- Initiate MLModel ---")
+
+model = model_factory(build_model_params["model_type"])
+
+logger.info(f"--- Initiate DataClient ---")
+
+# Use data client for input and output, as we use dvc to cache later to the cloud
+input_dataclient_type = scenarios_params["input_dataclient_type"]
+input_dataclient = dataclient_factory(
+    dataclient_type=input_dataclient_type,
+    dataclient_config=client_params[input_dataclient_type],
+)
+
+output_dataclient_type = scenarios_params["output_dataclient_type"]
+output_dataclient = dataclient_factory(
+    dataclient_type=output_dataclient_type,
+    dataclient_config=client_params[output_dataclient_type],
+)
+
+logger.info(f"--- Initiate MLMetrics ---")
+
+metrics = metrics_factory(generate_metrics_params["metrics_type"])
+
+
+def generate_scenario_predictions(
+    input_dataclient: DataClient,
+    output_dataclient: DataClient,
+    model: MLModel,
+    metrics: MLMetrics,
+    model_filepath: str,
+    scenario_data_filepaths: list,
+    predictions_column_name: str,
+    comparison_output_filepath: str,
+    metrics_output_filepath: str,
+):
+    """
+    Given the new model, we generate prediction for expected scenarios
+    """
+
+    logger.info("--- Loading Scenario Data ---")
+
+    scenario_data = pd.DataFrame()
+
+    # If we have no scenario data, we can save empty dataframes
+    if scenario_data_filepaths is None:
+        logger.info("No scenario data filepaths provided")
+        output_dataclient.save_data(
+            obj=scenario_data, location=comparison_output_filepath, save_config=None
+        )
+
+        output_dataclient.save_data(
+            obj=scenario_data, location=metrics_output_filepath, save_config=None
+        )
+        return
+
+    # Can have multiple scenario data files
+    for scenario_data_filepath in scenario_data_filepaths:
+        scenario_data = pd.concat(
+            [
+                scenario_data,
+                input_dataclient.load_data(scenario_data_filepath, load_config=None),
+            ]
+        )
+
+    logger.info("--- Loading Model ---")
+
+    model.load_model(model_filepath)
+
+    logger.info("--- Generating Predictions ---")
+
+    predictions = model.predict(
+        data=scenario_data, post_prediction_logic=post_prediction_logic
+    )
+
+    logger.info("--- Generate Scenario Predicted Impact ---")
+
+    predictions_df = pd.DataFrame(predictions)
+    predictions_df.columns = [predictions_column_name]
+
+    scenario_data = pd.concat([scenario_data, predictions_df], axis=1)
+    scenario_data["predicted_impact"] = abs(
+        scenario_data[predictions_column_name] - scenario_data["sap_starting"]
+    )
+
+    logger.info("--- Generate Metrics ---")
+
+    metrics_dict = metrics.generate_metrics(
+        scenario_data["impact"], scenario_data["predicted_impact"]
+    )
+
+    metrics_df = pd.DataFrame(metrics_dict, index=[0]).T.reset_index()
+    metrics_df.columns = ["metric", "value"]
+
+    logger.info("--- Save prediction into metrics ---")
+
+    output_df = scenario_data[["uprn", "id", "impact", "predicted_impact"]]
+
+    output_dataclient.save_data(
+        obj=output_df, location=comparison_output_filepath, save_config=None
+    )
+
+    output_dataclient.save_data(
+        obj=metrics_df, location=metrics_output_filepath, save_config=None
+    )
+
+
+if __name__ == "__main__":
+
+    logger.info(f"--- {__file__} - Start! ---")
+
+    logger.info(f"--- Generate Scenario Predictions ---")
+
+    generate_scenario_predictions(
+        input_dataclient=input_dataclient,
+        output_dataclient=output_dataclient,
+        model=model,
+        metrics=metrics,
+        model_filepath=model_filepath,
+        scenario_data_filepaths=scenario_data_filepaths,
+        predictions_column_name=predictions_column_name,
+        comparison_output_filepath=comparison_output_filepath,
+        metrics_output_filepath=metrics_output_filepath,
+    )
+
+    logger.info(f"--- {__file__} - Complete! ---")

modules/ml-pipeline/src/pipeline/README.md

@@ -37,3 +37,4 @@ Workflow:
     - This experiment will have the corresponding .dvc files for the hashed model and data
 - Use version control as normal
     - git add, git commit etc
+- To revert change, use `git checkout {COMMIT_HASH}`, followed by `git switch -c {NEW_BRANCH_NAME}`

modules/ml-pipeline/src/pipeline/config.py

@@ -7,6 +7,7 @@ settings = Dynaconf(
         "./configs/settings.yaml",
         "./configs/build_model.yaml",
         "./configs/analysis.yaml",
+        "./configs/scenarios.yaml",
     ],
 )
 

modules/ml-pipeline/src/pipeline/configs/analysis.yaml

@@ -13,4 +13,4 @@ default:
     dataclient_type: local
     nshap_samples: 100 # how many samples to use to approximate each Shapely value, larger values will be slower
     n_val: 30  # how many datapoints from validation data should we interpret predictions for, larger values will be slower
-    row_index: [0, 10, 20] # index of an example datapoint
+    row_index: [20695, 50243, 7653] # index of an example datapoint

modules/ml-pipeline/src/pipeline/configs/build_model.yaml

@@ -1,8 +1,9 @@
 default:
   build_model:
     model_type: AutogluonAutoML
-    model_save_filepath: ./data/model/autogluonmodel/
+    model_save_filepath: ./data/model/optimised/
     fit_metrics_filepath: ./metrics/fit_metrics.json
+    fit_predictions_filepath: ./data/fit_predictions/predictions.parquet
 
     SKLearnLinearRegression: null
 
@@ -10,9 +11,12 @@ default:
       kernel: "linear"
 
     AutogluonAutoML:
-      output_filepath: ./data/model/autogluonmodel/
+      output_filepath: ./data/model/allmodels/
       problem_type: regression
-      eval_metric: mean_absolute_error
-      time_limit: 75
+      eval_metric: mean_squared_error #mean_absolute_error
+      time_limit: 1800
       presets: medium_quality
-      excluded_model_types: ['KNN']
+      excluded_model_types: ['RF', 'CAT', 'NN_TORCH', 'KNN', 'XT']
+      infer_limit: 0.05
+      infer_limit_batch_size: 10000
+      ag_args_ensemble: {'num_folds_parallel': 2}

modules/ml-pipeline/src/pipeline/configs/feature_processor_logic.py

@@ -9,15 +9,42 @@ Business Logic dict + functions
 
 def remove_starting_columns(df):
     keep_column_index = [
-        False if col_name.endswith("_STARTING") else True
+        False if col_name.endswith("_starting") else True
         for col_name in list(df.columns)
     ]
     keep_columns = df.columns[keep_column_index].to_list()
-    keep_columns.append("SAP_STARTING")
+    keep_columns.append("sap_starting")
     df = df[keep_columns]
     return df
 
 
+def remove_floor_height_ending(df):
+    # df.describe(percentiles=[0.005,0.99])['FLOOR_HEIGHT_ENDING']
+    # shows bottom 0.5 percentile is 1.665
+    # So keep anything above this
+    df = df[df["floor_height_ending"] > 1.665].reset_index(drop=True)
+    print("we in here")
+    return df
+
+
+def remove_minimum_habitable_room_size(df):
+    # Need minimum of 6.5m per habitable room
+    df = df[
+        df["total_floor_area_ending"] / df["number_habitable_rooms"] > 6.5
+    ].reset_index(drop=True)
+    return df
+
+
+def keep_flats(df):
+    df = df[df["property_type"] == "Flat"]
+    return df
+
+
+def keep_non_zero_rdsap(df):
+    df = df[df["rdsap_change"] != 0]
+    return df
+
+
 # def keep_ending_columns(df):
 #     ending_column_index = [ col_name.endswith("_ENDING") for col_name in list(df.columns)]
 #     keep_columns = df.columns[ending_column_index].to_list()
@@ -27,7 +54,11 @@ def remove_starting_columns(df):
 #     return df
 
 business_logic = {
-    "remove_starting_columns": remove_starting_columns
+    # "keep_non_zero_rdsap": keep_non_zero_rdsap,
+    # "keep_flats": keep_flats,
+    # "remove_minimum_habitable_room_size": remove_minimum_habitable_room_size,
+    # "remove_floor_height_ending": remove_floor_height_ending
+    # "remove_starting_columns": remove_starting_columns
     # "keep_ENDING_COLUMNS": keep_ending_columns
 }
 

modules/ml-pipeline/src/pipeline/configs/post_prediction_logic.py

@@ -5,16 +5,18 @@ import pandas as pd
 
 
 def clip_predictions_to_minimum_value(
-    data: pd.DataFrame, predictions: pd.Series, minimum_value: int = 1
+    data: pd.DataFrame, predictions: pd.Series, minimum_value: int = 0
 ) -> pd.Series:
 
     series_name = predictions.name
     predictions.name = "predictions"
     predictions_df = pd.concat([data, predictions], axis=1)
     # We expect all prediction to be atleast one point improvement
-    replace_index = predictions_df["SAP_STARTING"] + 1 > predictions_df["predictions"]
+    replace_index = (
+        predictions_df["sap_starting"] + minimum_value > predictions_df["predictions"]
+    )
     predictions_df.loc[replace_index, "predictions"] = (
-        predictions_df.loc[replace_index, "SAP_STARTING"] + minimum_value
+        predictions_df.loc[replace_index, "sap_starting"] + minimum_value
     )
 
     predictions_new = predictions_df["predictions"]

modules/ml-pipeline/src/pipeline/configs/scenarios.yaml

@@ -0,0 +1,13 @@
+default:
+  scenarios:
+    input_dataclient_type: aws-s3
+    output_dataclient_type: local
+    scenario_data_filepaths:
+      # - s3://retrofit-data-dev/scenario_data/22-03-2024-19-20-09/recommendations_scoring_data.parquet
+      # - s3://retrofit-data-dev/scenario_data/24-03-2024-20-23-25/recommendations_scoring_data.parquet
+      # - s3://retrofit-data-dev/scenario_data/27-03-2024-11-38-15/recommendations_scoring_data.parquet
+      # - s3://retrofit-data-dev/scenario_data/26-05-2024-08-47-45/recommendations_scoring_data.parquet
+      # - s3://retrofit-data-dev/scenario_data/26-05-2024-10-44-53/recommendations_scoring_data.parquet
+      - s3://retrofit-data-dev/scenario_data/28-05-2024-19-22-41/recommendations_scoring_data.parquet
+    comparison_output_filepath: ./metrics/scenario_table.md
+    metrics_output_filepath: ./metrics/scenario_metrics.md

modules/ml-pipeline/src/pipeline/configs/settings.yaml

@@ -18,7 +18,10 @@ default:
   prepare_data:
     input_dataclient_type: aws-s3
     output_dataclient_type: local
-    data_filepath: s3://retrofit-data-dev/sap_change_model/dataset_without_differencing.parquet
+    # data_filepath: s3://retrofit-data-dev/sap_change_model/2024-03-22-18-56-53/dataset_rooms.parquet
+    # data_filepath: s3://retrofit-data-dev/sap_change_model/2024-05-25-08-36-36/dataset_rooms.parquet
+    # data_filepath: s3://retrofit-data-dev/sap_change_model/2024-05-26-10-31-39/dataset_rooms.parquet
+    data_filepath: s3://retrofit-data-dev/sap_change_model/2024-05-28-19-08-25/dataset_rooms.parquet
     train_proportion: 0.9
     output_train_filepath: ./data/prepared_data/train.parquet
     output_test_filepath: ./data/prepared_data/test.parquet
@@ -28,9 +31,37 @@ default:
     feature_processor_config:
       subsample_amount: null
       subsample_seed: 0
-      target: SAP_ENDING
-      drop_columns: ["UPRN", "HEAT_DEMAND_CHANGE", "CARBON_CHANGE", "RDSAP_CHANGE", "HEAT_DEMAND_ENDING", "CARBON_ENDING"]
+      target: sap_ending
+      identifier_columns: ["uprn"]
+      # drop_columns: ["heat_demand_change", "carbon_change", "rdsap_change", "heat_demand_ending", "carbon_ending", "days_to_starting", "days_to_ending"]
+      drop_columns: [
+        "heat_demand_change", "carbon_change", "rdsap_change", "heat_demand_ending", "carbon_ending", "days_to_starting", "days_to_ending",
+        'number_habitable_rooms_starting', 'number_habitable_rooms_ending', 'number_heated_rooms_starting', 'number_heated_rooms_ending',
+        'number_habitable_rooms', 'number_heated_rooms']
       retain_features: null
+      # retain_features: ['uprn', 'sap_starting', 'hot_water_energy_eff_ending',
+      #  'mainheat_energy_eff_ending', 'constituency', 'roof_energy_eff_ending',
+      #  'walls_energy_eff_ending', 'secondheat_description_ending',
+      #  'property_type', 'mainheatc_energy_eff_ending', 'built_form',
+      #  'walls_insulation_thickness_ending', 'potential_energy_efficiency',
+      #  'transaction_type_ending',
+      #  'floor_thermal_transmittance_ending',
+      #  'low_energy_lighting_ending', 'heat_demand_starting',
+      #  'photo_supply_ending', 'carbon_starting',
+      #  'walls_thermal_transmittance_ending',
+      #  'roof_insulation_thickness_ending',
+      #  'total_floor_area_ending', 'number_open_fireplaces_ending',
+      #  'windows_energy_eff_ending',
+      #  'floor_height_ending',
+      #  'extension_count_ending',
+      #  'has_air_source_heat_pump_ending',
+      #  'charging_system_ending', 'construction_age_band', 'glazed_type_ending',
+      #  'roof_thermal_transmittance_ending',
+      #  'floor_insulation_thickness_ending', 'has_mains_gas_ending',
+      #  'estimated_perimeter_starting', 'energy_consumption_potential',
+      #  'environment_impact_potential', 'heater_type_ending',
+      #  'multi_glaze_proportion_ending',
+      #  'lighting_energy_eff_ending', 'fixed_lighting_outlets_count']
 
   generate_predictions:
     input_dataclient_type: local
@@ -38,6 +69,7 @@ default:
     test_data_filepath: ./data/prepared_data/test.parquet
     predictions_output_filepath: ./data/predictions/predictions.parquet
     predictions_column_name: predictions
+    identifier_column: id
 
   generate_metrics:
     dataclient_type: local

modules/ml-pipeline/src/pipeline/core/DataClient.py

@@ -142,9 +142,15 @@ class AWSS3Client:
         buffer = BytesIO()
         obj.to_parquet(buffer, index=False)
 
+        # Reset the buffer position to the beginning
+        buffer.seek(0)
+
         bucket, key = location.strip("s3://").split("/", 1)
         self.client.upload_fileobj(buffer, bucket, key)
 
+        # Close the buffer
+        buffer.close()
+
     def _load_parquet(self, location: str, load_config: dict) -> pd.DataFrame:
         """
         Load a parquet file
@@ -239,7 +245,8 @@ class LocalClient:
 
         save_methods = {
             ".parquet": self._save_parquet,
-            ".json": self._save_json
+            ".json": self._save_json,
+            ".md": self._save_md,
             # "": _save_directory(**save_config),
             # ADD MORE save_methods HERE
         }
@@ -288,3 +295,10 @@ class LocalClient:
         # Write the contents of the buffer to the local file
         with open(location, "wb") as f:
             f.write(buffer.getvalue())
+
+    def _save_md(self, obj: pd.DataFrame, location: str, save_config: dict):
+        """
+        Save object as markdown
+        """
+
+        obj.to_markdown(location, **save_config)

modules/ml-pipeline/src/pipeline/core/Logger.py

@@ -21,6 +21,7 @@ def setup_logger():
 
     # Add the stream handler to the logger
     logger.addHandler(stream_handler)
+    logger.propagate = False
 
     return logger
 

modules/ml-pipeline/src/pipeline/core/MLMetrics.py

@@ -4,6 +4,7 @@ Implementation of MLMetrics, all of which will have two methods:
 - Generate Plot Suite
 """
 
+import numpy as np
 import pandas as pd
 from typing import Union
 from sklearn.metrics import (
@@ -14,6 +15,18 @@ from sklearn.metrics import (
 )
 from core.interface.InterfaceMetrics import MLMetrics
 
+# Define the function to return the SMAPE value
+def symmetric_mape(actual, predicted) -> float:
+
+    # Convert actual and predicted to numpy
+    # array data type if not already
+    if not all([isinstance(actual, np.ndarray), isinstance(predicted, np.ndarray)]):
+        actual, predicted = np.array(actual), np.array(predicted)
+
+    return np.mean(
+        np.abs(predicted - actual) / ((np.abs(predicted) + np.abs(actual)) / 2)
+    )
+
 
 def metrics_factory(metrics_type: str) -> MLMetrics:
     metrics = {
@@ -34,7 +47,7 @@ class RegressionMetrics:
         median_absolute_error,
         mean_squared_error,
         mean_absolute_percentage_error,
-        # max_error
+        symmetric_mape,
     ]
 
     def generate_metrics(

modules/ml-pipeline/src/pipeline/core/MLModels.py

@@ -25,7 +25,7 @@ def model_factory(model_type: str) -> MLModel:
     models = {
         "SKLearnLinearRegression": SKLearnLinearRegression(),
         "SKLearnSVMRegression": SKLearnSVMRegression(),
-        "AutogluonAutoML": AutogluonAutoML()
+        "AutogluonAutoML": AutogluonAutoML(),
         # ADD OTHER MODELS HERE
     }
 
@@ -149,6 +149,9 @@ class AutogluonAutoML:
         "time_limit",
         "presets",
         "excluded_model_types",
+        "infer_limit",
+        "infer_limit_batch_size",
+        "ag_args_ensemble",
     ]
 
     def load_model(self, path: Union[Path, str]) -> None:
@@ -165,8 +168,12 @@ class AutogluonAutoML:
         if self.model is None:
             raise KeyError("No model trained/ loaded - unable to save")
 
-        logger.info("In local development mode - no need for s3 client")
-        logger.info("Using AutoGluon Model - Model saving already occured")
+        logger.info(
+            "Using AutoGluon Model - Model saving is using optimised deployment mode"
+        )
+
+        logger.info("Saving optimised model")
+        self.model.clone_for_deployment(str(path))
 
         return str(path)
 
@@ -199,6 +206,9 @@ class AutogluonAutoML:
             time_limit=model_hyperparameters["time_limit"],
             presets=model_hyperparameters["presets"],
             excluded_model_types=model_hyperparameters["excluded_model_types"],
+            infer_limit=model_hyperparameters["infer_limit"],
+            infer_limit_batch_size=model_hyperparameters["infer_limit_batch_size"],
+            ag_args_ensemble=model_hyperparameters["ag_args_ensemble"],
         )
 
     def predict(

modules/ml-pipeline/src/pipeline/data/.gitignore

@@ -1,3 +0,0 @@
-/prepared_data
-/model
-/predictions

modules/ml-pipeline/src/pipeline/dvc.lock

@@ -1,27 +1,46 @@
 schema: '2.0'
 stages:
+  startup_cleanup:
+    cmd: python 0_startup_cleanup.py
+    deps:
+    - path: 0_startup_cleanup.py
+      hash: md5
+      md5: b1b12f6b6393fbf8b83d23684df0a3d4
+      size: 1220
+    params:
+      configs/settings.yaml:
+        default.startup_cleanup.artefacts: ./data
+        default.startup_cleanup.metrics: ./metrics
   prepare_data:
     cmd: python 1_prepare_data.py
     deps:
     - path: 1_prepare_data.py
       hash: md5
-      md5: c9f030df733e318b80d1fa91b7732f79
-      size: 5132
+      md5: 11a3b8bfdfe199ab7ecc39ccc5652649
+      size: 4298
     params:
       configs/settings.yaml:
         default.feature_processor.feature_processor_config.drop_columns:
-        - UPRN
-        - HEAT_DEMAND_CHANGE
-        - CARBON_CHANGE
-        - RDSAP_CHANGE
-        - HEAT_DEMAND_ENDING
-        - CARBON_ENDING
+        - heat_demand_change
+        - carbon_change
+        - rdsap_change
+        - heat_demand_ending
+        - carbon_ending
+        - days_to_starting
+        - days_to_ending
+        - number_habitable_rooms_starting
+        - number_habitable_rooms_ending
+        - number_heated_rooms_starting
+        - number_heated_rooms_ending
+        - number_habitable_rooms
+        - number_heated_rooms
         default.feature_processor.feature_processor_config.retain_features:
         default.feature_processor.feature_processor_config.subsample_amount:
         default.feature_processor.feature_processor_config.subsample_seed: 0
-        default.feature_processor.feature_processor_config.target: SAP_ENDING
+        default.feature_processor.feature_processor_config.target: sap_ending
         default.feature_processor.feature_processor_type: dataframe
-        default.prepare_data.data_filepath: s3://retrofit-data-dev/sap_change_model/dataset_without_differencing.parquet
+        default.prepare_data.data_filepath:
+          s3://retrofit-data-dev/sap_change_model/2024-05-28-19-08-25/dataset_rooms.parquet
         default.prepare_data.input_dataclient_type: aws-s3
         default.prepare_data.output_dataclient_type: local
         default.prepare_data.output_test_filepath: ./data/prepared_data/test.parquet
@@ -30,65 +49,79 @@ stages:
     outs:
     - path: data/prepared_data/
       hash: md5
-      md5: ed19a11a85d6a2006631173f51569d27.dir
-      size: 21131576
+      md5: 80c9e138146a1d96b9d16091c207e2e8.dir
+      size: 45056059
       nfiles: 2
   build_model:
     cmd: python 2_build_model.py
     deps:
     - path: 2_build_model.py
       hash: md5
-      md5: 039578b629d7cd204016e92cd079ea90
-      size: 5181
+      md5: 7231450b78920b0c5e7c6bada496b24a
+      size: 4820
     - path: data/prepared_data
       hash: md5
-      md5: ed19a11a85d6a2006631173f51569d27.dir
-      size: 21131576
+      md5: 80c9e138146a1d96b9d16091c207e2e8.dir
+      size: 45056059
       nfiles: 2
     params:
       configs/build_model.yaml:
         default:
           build_model:
             model_type: AutogluonAutoML
-            model_save_filepath: ./data/model/autogluonmodel/
+            model_save_filepath: ./data/model/optimised/
             fit_metrics_filepath: ./metrics/fit_metrics.json
+            fit_predictions_filepath: ./data/fit_predictions/predictions.parquet
             SKLearnLinearRegression:
             SKLearnSVMRegression:
               kernel: linear
             AutogluonAutoML:
-              output_filepath: ./data/model/autogluonmodel/
+              output_filepath: ./data/model/allmodels/
               problem_type: regression
-              eval_metric: mean_absolute_error
-              time_limit: 75
+              eval_metric: mean_squared_error
+              time_limit: 1800
               presets: medium_quality
               excluded_model_types:
+              - RF
+              - CAT
+              - NN_TORCH
               - KNN
+              - XT
+              infer_limit: 0.05
+              infer_limit_batch_size: 10000
+              ag_args_ensemble:
+                num_folds_parallel: 2
     outs:
+    - path: data/fit_predictions/
+      hash: md5
+      md5: d9c9afc05e8780db47c0548b19bf7d19.dir
+      size: 3349989
+      nfiles: 1
     - path: data/model/
       hash: md5
-      md5: 60e253c42cc36934098c627ef3ef4cc1.dir
-      size: 185134993
-      nfiles: 14
+      md5: 13c3100e1486c27a83a8a47491077842.dir
+      size: 773523079
+      nfiles: 36
     - path: metrics/fit_metrics.json
       hash: md5
-      md5: a0c2a1c9e5da0b857d510fa1ba6282a8
-      size: 186
+      md5: 2ff70a2a45813e1bcdf2ea3aa8e07d4a
+      size: 224
   generate_predictions:
     cmd: python 3_generate_predictions.py
     deps:
     - path: 3_generate_predictions.py
       hash: md5
-      md5: 238b3fa9f3c6f3720e77c116857070ae
-      size: 4720
+      md5: 0a70ad4dfe99414a75d1261c75a177b9
+      size: 2464
     - path: data/model
       hash: md5
-      md5: 60e253c42cc36934098c627ef3ef4cc1.dir
-      size: 185134993
-      nfiles: 14
+      md5: 13c3100e1486c27a83a8a47491077842.dir
+      size: 773523079
+      nfiles: 36
     - path: data/prepared_data
       hash: md5
-      md5: ed19a11a85d6a2006631173f51569d27.dir
-      size: 21131576
+      md5: 80c9e138146a1d96b9d16091c207e2e8.dir
+      size: 45056059
       nfiles: 2
     params:
       configs/settings.yaml:
@@ -100,25 +133,25 @@ stages:
     outs:
     - path: data/predictions/
       hash: md5
-      md5: 700c8767de3a86c4c5339daf3cc17192.dir
-      size: 380962
+      md5: 5d07bcebf3160a72bb18dfd79106e85c.dir
+      size: 463197
       nfiles: 1
   generate_metrics:
     cmd: python 4_generate_metrics.py
     deps:
     - path: 4_generate_metrics.py
       hash: md5
-      md5: 2c9fb78955a8c19cff0a098976f81d1b
-      size: 4487
+      md5: 4fedb86d89d528f0a6597934ba3890a0
+      size: 3484
     - path: data/predictions
       hash: md5
-      md5: 700c8767de3a86c4c5339daf3cc17192.dir
-      size: 380962
+      md5: 5d07bcebf3160a72bb18dfd79106e85c.dir
+      size: 463197
       nfiles: 1
     - path: data/prepared_data
       hash: md5
-      md5: ed19a11a85d6a2006631173f51569d27.dir
-      size: 21131576
+      md5: 80c9e138146a1d96b9d16091c207e2e8.dir
+      size: 45056059
       nfiles: 2
     params:
       configs/settings.yaml:
@@ -128,16 +161,30 @@ stages:
     outs:
     - path: metrics/metrics.json
       hash: md5
-      md5: 45ffac8f6e7283df4b69af8a9abc45e1
-      size: 184
-  startup_cleanup:
-    cmd: python 0_startup_cleanup.py
+      md5: 3e08df02fd5c5d094bcf936e1338d596
+      size: 223
+  generate_scenerio_metrics:
+    cmd: python 5_generate_scenarios.py
     deps:
-    - path: 0_startup_cleanup.py
+    - path: 5_generate_scenarios.py
       hash: md5
-      md5: fbb7e3b1b98b517c870f3e1df3e7f695
-      size: 1676
+      md5: 40506749fefd926d47c60ff5b16db307
+      size: 5337
     params:
-      configs/settings.yaml:
-        default.startup_cleanup.artefacts: ./data
-        default.startup_cleanup.metrics: ./metrics
+      configs/scenarios.yaml:
+        default.scenarios:
+          input_dataclient_type: aws-s3
+          output_dataclient_type: local
+          scenario_data_filepaths:
+          - s3://retrofit-data-dev/scenario_data/28-05-2024-19-22-41/recommendations_scoring_data.parquet
+          comparison_output_filepath: ./metrics/scenario_table.md
+          metrics_output_filepath: ./metrics/scenario_metrics.md
+    outs:
+    - path: metrics/scenario_metrics.md
+      hash: md5
+      md5: fa4d6d7bbd7818613800da5f8f37ea96
+      size: 363
+    - path: metrics/scenario_table.md
+      hash: md5
+      md5: d6baf100a1623cc2467c2f8221d314c9
+      size: 2133

modules/ml-pipeline/src/pipeline/dvc.yaml

@@ -38,6 +38,7 @@ stages:
     - configs/build_model.yaml:
     outs:
     - data/model/
+    - data/fit_predictions/
     - metrics/fit_metrics.json
     always_changed: true
   generate_predictions:
@@ -70,6 +71,17 @@ stages:
     outs:
     - metrics/metrics.json
     always_changed: true
+  generate_scenerio_metrics:
+    cmd: python 5_generate_scenarios.py
+    deps:
+    - 5_generate_scenarios.py
+    params:
+    - configs/scenarios.yaml:
+      - default.scenarios
+    outs:
+    - metrics/scenario_table.md
+    - metrics/scenario_metrics.md
+    always_changed: true
 metrics:
   - metrics/metrics.json
   - metrics/fit_metrics.json

modules/ml-pipeline/src/pipeline/eda.py

@@ -175,3 +175,74 @@ plot_permutation_importance(exp, fig_kw={"figwidth": 7, "figheight": 6})
 # Use shap package to explain why 9158 has a 35 prediction when its sap ending is 96
 #
 #
+
+from core.MLModels import model_factory
+from core.DataClient import dataclient_factory
+import pandas as pd
+from config import settings
+
+client_params = settings.client
+prepare_data_params = settings.prepare_data
+feature_process_params = settings.feature_processor
+build_model_params = settings.build_model
+generate_predictions_params = settings.generate_predictions
+prediction_analysis_params = settings.prediction_analysis
+model = model_factory(build_model_params["model_type"])
+model.load_model(build_model_params["model_save_filepath"])
+dataclient_type = prediction_analysis_params["dataclient_type"]
+# dataclient_type = 'aws-s3'
+# dataclient = dataclient_factory(
+#     dataclient_type=dataclient_type,
+#     dataclient_config=client_params[dataclient_type],
+# )
+# data = dataclient.load_data("s3://retrofit-data-dev/sap_change_model/dataset.parquet")
+
+target = feature_process_params["feature_processor_config"]["target"]
+predictions_column_name = generate_predictions_params["predictions_column_name"]
+output_test_filepath = prepare_data_params["output_test_filepath"]
+predictions_output_filepath = generate_predictions_params["predictions_output_filepath"]
+
+# score_data = dataclient.load_data("s3://retrofit-data-dev/carbon_change_predictions/51/2023-11-28T21:01:21.869339.parquet")
+
+
+local_dataclient = dataclient_factory(
+    dataclient_type="local",
+    dataclient_config=client_params["local"],
+)
+test_df = local_dataclient.load_data(output_test_filepath)
+predictions = local_dataclient.load_data(predictions_output_filepath)
+mix_df = pd.concat([test_df.copy(), predictions], axis=1)
+mix_df["residual"] = abs(mix_df[predictions_column_name] - mix_df[target])
+mix_df = mix_df.sort_values("residual", ascending=False)
+
+cosine_similarity_df = mix_df[mix_df.columns.difference(["predictions", "residual"])]
+from sklearn.metrics.pairwise import cosine_similarity
+
+row_index = 0
+
+from sklearn.preprocessing import LabelEncoder
+
+object_columns = cosine_similarity_df.select_dtypes(["object"])
+
+cosine_similarity_df[object_columns.columns] = cosine_similarity_df[
+    object_columns.columns
+].apply(LabelEncoder().fit_transform)
+
+feature_vector = cosine_similarity_df.loc[[row_index]]
+
+cosine_similarity_df["cosine"] = cosine_similarity(cosine_similarity_df, feature_vector)
+similar_index = (
+    cosine_similarity_df.sort_values("cosine", ascending=False).head(15).index
+)
+
+check_df = mix_df.loc[similar_index]
+
+columns_to_check = [
+    "LOW_ENERGY_LIGHTING_ENDING",
+    "walls_thermal_transmittance_ENDING",
+    "floor_thermal_transmittance_ENDING",
+    "roof_thermal_transmittance_ENDING",
+    "roof_insulation_thickness_ENDING",
+]
+
+cosine_similarity_df = mix_df[columns_to_check]

modules/ml-pipeline/src/pipeline/generate_predictions.py

@@ -0,0 +1,56 @@
+import pandas as pd
+from configs.post_prediction_logic import post_prediction_logic
+from core.interface.InterfaceModels import MLModel
+from core.interface.InterfaceDataClient import DataClient
+from core.Logger import logger
+
+
+def generate_predictions(
+    input_dataclient: DataClient,
+    output_dataclient: DataClient,
+    model: MLModel,
+    target: str,
+    model_filepath: str,
+    test_data_filepath: str,
+    predictions_output_filepath: str,
+    predictions_column_name: str,
+    identifier_column: str = "id",
+):
+    """
+    For a given model, we generate prediction and evaluate this against the true target
+    """
+
+    logger.info("--- Loading test data ---")
+
+    test_data = input_dataclient.load_data(
+        location=test_data_filepath, load_config=None
+    )
+
+    logger.info("--- Loading model ---")
+
+    model.load_model(model_filepath)
+
+    logger.info("--- Generating predictions ---")
+
+    prediction_data = (
+        test_data.drop(columns=target) if target in test_data.columns else test_data
+    )
+
+    predictions = model.predict(
+        data=prediction_data, post_prediction_logic=post_prediction_logic
+    )
+
+    logger.info("--- Saving predictions ---")
+
+    predictions_df = pd.DataFrame(predictions)
+    predictions_df.columns = [predictions_column_name]
+
+    output_df = (
+        pd.concat([test_data[identifier_column], predictions_df], axis=1)
+        if identifier_column in test_data.columns
+        else predictions_df
+    )
+
+    output_dataclient.save_data(
+        obj=output_df, location=predictions_output_filepath, save_config=None
+    )

modules/ml-pipeline/src/pipeline/metrics/.gitignore

@@ -1,2 +1,4 @@
 /fit_metrics.json
 /metrics.json
+/scenario_table.md
+/scenario_metrics.md

modules/ml-pipeline/src/pipeline/requirements/predictions/requirements-dev.txt

@@ -1,7 +1,7 @@
 joblib==1.3.2
 boto3==1.28.17
-pandas==1.5.3
-autogluon==0.8.2
-dynaconf==3.2.0
+pandas==2.1.4
+autogluon.tabular[all]==1.0.0
+dynaconf==3.2.1
 pyarrow==13.0.0
 pre-commit==3.3.3

modules/ml-pipeline/src/pipeline/requirements/predictions/requirements.txt

@@ -1,7 +1,7 @@
 joblib==1.3.2
 boto3==1.28.17
-pandas==1.5.3
-autogluon==0.8.2
-dynaconf==3.2.0
+pandas==2.1.4
+autogluon.tabular[all]==1.0.0
+dynaconf==3.2.1
 pyarrow==13.0.0
 PyYAML==6.0.1

modules/ml-pipeline/src/pipeline/requirements/training/requirements-dev.txt

@@ -1,9 +1,10 @@
 joblib==1.3.2
 boto3==1.28.17
-pandas==1.5.3
-autogluon==0.8.2
-dynaconf==3.2.0
-alibi==0.9.4
+pandas==2.1.4
+autogluon.tabular[all]==1.0.0
+ray==2.6.3
+dynaconf==3.2.1
+alibi==0.9.5
 shap==0.42.1
 pyarrow==13.0.0
 pre-commit==3.3.3

modules/ml-pipeline/src/pipeline/requirements/training/requirements.txt

@@ -1,4 +1,4 @@
 boto3==1.28.41
-pandas==1.5.3
-autogluon==0.8.2
-dynaconf==3.2.0
+pandas==2.1.4
+autogluon.tabular[all]==1.0.0
+dynaconf==3.2.1

modules/ml-pipeline/src/pipeline/requirements/version_control/requirements.txt

@@ -1,3 +1,4 @@
-dvc==3.18.0
-dvc-s3==2.23.0
-gto==1.0.4
+dvc==3.51.0
+dvc-s3==3.2.0
+gto==1.7.1
+pyOpenSSL==23.3.0