Merge pull request #87 from Hestia-Homes/heat-dev-model

add restriction to datast

MODEL_REGISTRY.md

@@ -8,6 +8,14 @@
         "active": true
     },
     "sap": {
+        "version": "v0.2.6",
+        "stage": {
+            "dev": "v0.2.6"
+        },
+        "registered": true,
+        "active": true
+    },
+    "heat": {
         "version": "v0.1.0",
         "stage": {
             "dev": "v0.1.0"
@@ -15,10 +23,10 @@
         "registered": true,
         "active": true
     },
-    "heat": {
+    "carbon": {
-        "version": "v0.0.1",
+        "version": "v0.1.0",
         "stage": {
-            "dev": "v0.0.1"
+            "dev": "v0.1.0"
         },
         "registered": true,
         "active": true

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

@@ -4,9 +4,7 @@ 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.InterfaceMetrics import MLMetrics
 from core.interface.InterfaceDataClient import DataClient

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

@@ -13,7 +13,7 @@ default:
       output_filepath: ./data/model/allmodels/
       problem_type: regression
       eval_metric: mean_squared_error #mean_absolute_error
-      time_limit: 4000
+      time_limit: 400
       presets: medium_quality
       excluded_model_types: ['KNN', 'RF']
       infer_limit: 0.05

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

@@ -18,6 +18,42 @@ def remove_starting_columns(df):
     return df
 
 
+def keep_negative_heat_change(df):
+    df = df[df["HEAT_DEMAND_CHANGE"] < 0]
+    return df
+
+
+def keep_negative_carbon_change(df):
+    df = df[df["CARBON_CHANGE"] < 0]
+    return df
+
+
+# TODO: Move to ETL pipeline
+def remove_unreasonable_habitable_rooms(df):
+    """
+    Assumption is that proportion of floor area to habitable rooms should be at least 6.5m2
+    """
+    minimum_room_size_index = (
+        df["TOTAL_FLOOR_AREA_ENDING"] / df["NUMBER_HABITABLE_ROOMS"] >= 6.5
+    )
+    df = df[minimum_room_size_index]
+    return df
+
+
+def remove_top_1_percent_heat_demand(df):
+    # threshold_value = df.describe(percentiles=[0.99])['HEAT_DEMAND_STARTING']['99%']
+    threshold_value = 860
+    df = df[df["HEAT_DEMAND_STARTING"] < threshold_value]
+    return df
+
+
+def remove_top_1_percent_carbon(df):
+    # threshold_value = df.describe(percentiles=[0.99])['CARBON_STARTING']['99%']
+    threshold_value = 18
+    df = df[df["CARBON_STARTING"] < threshold_value]
+    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()
@@ -27,6 +63,11 @@ def remove_starting_columns(df):
 #     return df
 
 business_logic = {
+    "remove_unreasonable_habitable_rooms": remove_unreasonable_habitable_rooms,
+    "keep_negative_heat_change": keep_negative_heat_change,
+    "keep_negative_carbon_change": keep_negative_carbon_change,
+    "remove_top_1_percent_heat_demand": remove_top_1_percent_heat_demand,
+    "remove_top_1_percent_carbon": remove_top_1_percent_carbon,
     # "remove_starting_columns": remove_starting_columns
     # "keep_ENDING_COLUMNS": keep_ending_columns
 }

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

@@ -12,9 +12,11 @@ def clip_predictions_to_minimum_value(
     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["predictions"] > predictions_df["HEAT_DEMAND_STARTING"] - 1
+    )
     predictions_df.loc[replace_index, "predictions"] = (
-        predictions_df.loc[replace_index, "SAP_STARTING"] + minimum_value
+        predictions_df.loc[replace_index, "HEAT_DEMAND_STARTING"] - minimum_value
     )
 
     predictions_new = predictions_df["predictions"]

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

@@ -31,9 +31,9 @@ default:
     feature_processor_config:
       subsample_amount: null
       subsample_seed: 0
-      target: SAP_ENDING
+      target: HEAT_DEMAND_ENDING
       identifier_columns: ["UPRN"]
-      drop_columns: ["HEAT_DEMAND_CHANGE", "CARBON_CHANGE", "RDSAP_CHANGE", "HEAT_DEMAND_ENDING", "CARBON_ENDING"]
+      drop_columns: ["HEAT_DEMAND_CHANGE", "CARBON_CHANGE", "RDSAP_CHANGE", "SAP_ENDING", "CARBON_ENDING"]
       # retain_features: ["SAP_STARTING", "TOTAL_FLOOR_AREA_DIFF"]
       retain_features: null
 

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

@@ -5,22 +5,22 @@ stages:
     deps:
     - path: 1_prepare_data.py
       hash: md5
-      md5: c9f030df733e318b80d1fa91b7732f79
+      md5: 896d3d88a4a9f68d174efe71dc089517
-      size: 5132
+      size: 4222
     params:
       configs/settings.yaml:
         default.feature_processor.feature_processor_config.drop_columns:
         - HEAT_DEMAND_CHANGE
         - CARBON_CHANGE
         - RDSAP_CHANGE
-        - HEAT_DEMAND_ENDING
+        - SAP_ENDING
         - CARBON_ENDING
         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: HEAT_DEMAND_ENDING
         default.feature_processor.feature_processor_type: dataframe
-        default.prepare_data.data_filepath: s3://retrofit-data-dev/sap_change_model/dataset.parquet
+        default.prepare_data.data_filepath: s3://retrofit-data-dev/sap_change_model/dataset_test.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
@@ -29,20 +29,20 @@ stages:
     outs:
     - path: data/prepared_data/
       hash: md5
-      md5: 9ce5c45722da7fc40491b5a4d00daf9e.dir
+      md5: f235f38714fefcf6e4927ae95ba912c3.dir
-      size: 33881619
+      size: 30774760
       nfiles: 2
   build_model:
     cmd: python 2_build_model.py
     deps:
     - path: 2_build_model.py
       hash: md5
-      md5: 84699d208874c52accaff61c6af9bb0a
+      md5: b824822475c222521516493e68eef9c5
-      size: 5359
+      size: 4149
     - path: data/prepared_data
       hash: md5
-      md5: 9ce5c45722da7fc40491b5a4d00daf9e.dir
+      md5: f235f38714fefcf6e4927ae95ba912c3.dir
-      size: 33881619
+      size: 30774760
       nfiles: 2
     params:
       configs/build_model.yaml:
@@ -58,37 +58,39 @@ stages:
               output_filepath: ./data/model/allmodels/
               problem_type: regression
               eval_metric: mean_squared_error
-              time_limit: 4000
+              time_limit: 400
               presets: medium_quality
               excluded_model_types:
               - KNN
               - RF
+              infer_limit: 0.05
+              infer_limit_batch_size: 10000
     outs:
     - path: data/model/
       hash: md5
-      md5: 7bb5156243b4db39349e80a01ffecde4.dir
+      md5: a868845999b46e0272dc27f5cb5bc618.dir
-      size: 473398662
+      size: 310555147
-      nfiles: 27
+      nfiles: 24
     - path: metrics/fit_metrics.json
       hash: md5
-      md5: 2bb16ac67de8778fbc08171d562b34d5
+      md5: 809f27735c77cbcb62866b96018eedea
-      size: 184
+      size: 216
   generate_predictions:
     cmd: python 3_generate_predictions.py
     deps:
     - path: 3_generate_predictions.py
       hash: md5
-      md5: 5ef2856a5a977304f1ec01f9b4205262
+      md5: 0a70ad4dfe99414a75d1261c75a177b9
-      size: 3028
+      size: 2464
     - path: data/model
       hash: md5
-      md5: 7bb5156243b4db39349e80a01ffecde4.dir
+      md5: a868845999b46e0272dc27f5cb5bc618.dir
-      size: 473398662
+      size: 310555147
-      nfiles: 27
+      nfiles: 24
     - path: data/prepared_data
       hash: md5
-      md5: 9ce5c45722da7fc40491b5a4d00daf9e.dir
+      md5: f235f38714fefcf6e4927ae95ba912c3.dir
-      size: 33881619
+      size: 30774760
       nfiles: 2
     params:
       configs/settings.yaml:
@@ -100,25 +102,25 @@ stages:
     outs:
     - path: data/predictions/
       hash: md5
-      md5: 0bb3cf991906953def81c8204cdcfaf0.dir
+      md5: 2098fe82304751025e427f2cc241a2ff.dir
-      size: 374532
+      size: 295849
       nfiles: 1
   generate_metrics:
     cmd: python 4_generate_metrics.py
     deps:
     - path: 4_generate_metrics.py
       hash: md5
-      md5: 2c9fb78955a8c19cff0a098976f81d1b
+      md5: 567b1acb819e2ff432b989cdbdd4a2bf
-      size: 4487
+      size: 3448
     - path: data/predictions
       hash: md5
-      md5: 0bb3cf991906953def81c8204cdcfaf0.dir
+      md5: 2098fe82304751025e427f2cc241a2ff.dir
-      size: 374532
+      size: 295849
       nfiles: 1
     - path: data/prepared_data
       hash: md5
-      md5: 9ce5c45722da7fc40491b5a4d00daf9e.dir
+      md5: f235f38714fefcf6e4927ae95ba912c3.dir
-      size: 33881619
+      size: 30774760
       nfiles: 2
     params:
       configs/settings.yaml:
@@ -128,15 +130,15 @@ stages:
     outs:
     - path: metrics/metrics.json
       hash: md5
-      md5: 2e13ae67759a64261d03224f1c0d4bf4
+      md5: aa671878e1bd8c6a8d4b5f9788c817c4
-      size: 185
+      size: 219
   startup_cleanup:
     cmd: python 0_startup_cleanup.py
     deps:
     - path: 0_startup_cleanup.py
       hash: md5
-      md5: fbb7e3b1b98b517c870f3e1df3e7f695
+      md5: b1b12f6b6393fbf8b83d23684df0a3d4
-      size: 1676
+      size: 1220
     params:
       configs/settings.yaml:
         default.startup_cleanup.artefacts: ./data

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

@@ -38,7 +38,6 @@ train_df[[target, "SAP_STARTING"]].plot(y=target, x="SAP_STARTING", style="o")
 train_df[[target, "HEAT_DEMAND_STARTING"]].plot(
     x=target, y="HEAT_DEMAND_STARTING", style="o"
 )
-
 # Both make sense: i.e. the higher the sap, the lower we predict and the higher the heat demand, the higher we predict
 
 # Load the autogluon model and check feature importance
@@ -176,6 +175,8 @@ plot_permutation_importance(exp, fig_kw={"figwidth": 7, "figheight": 6})
 #
 #
 
+from core.MLMetrics import metrics_factory
+
 from core.MLModels import model_factory
 from core.DataClient import dataclient_factory
 import pandas as pd
@@ -206,6 +207,9 @@ 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)
 
+metrics = metrics_factory("Regression")
+metrics.generate_metrics(mix_df["predictions"], mix_df["HEAT_DEMAND_ENDING"])
+
 cosine_similarity_df = mix_df[
     mix_df.columns.difference(["predictions", "residual", "SAP_ENDING"])
 ]