new makefile

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

@@ -11,6 +11,6 @@ AutogluonAutoML:
   output_filepath: ./data/model/autogluonmodel/
   problem_type: regression
   eval_metric: mean_absolute_error
-  time_limit: 400
-  presets: good_quality
+  time_limit: 60
+  presets: medium_quality
   excluded_model_types: ['KNN']

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

@@ -17,3 +17,4 @@ def SAP_ENDING(df):
 
 
 new_feature_funcs = {"SAP_ENDING": SAP_ENDING}
+# new_feature_funcs = {}

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

@@ -5,8 +5,8 @@ stages:
     deps:
     - path: prepare_data.py
       hash: md5
-      md5: 934d774e67f38e440b621ce71152f5f6
-      size: 5031
+      md5: 2648d7d407dca857a1d20a11a88d3d98
+      size: 5116
     params:
       configs/prepare_data.yaml:
         output_test_filepath: ./data/prepared_data/test.parquet
@@ -15,8 +15,8 @@ stages:
     outs:
     - path: data/prepared_data/
       hash: md5
-      md5: 3767eec56906f5ac724a3f07433645ef.dir
-      size: 13442342
+      md5: ec064b0274e2e6a0864580a748e6bb6a.dir
+      size: 24067192
       nfiles: 2
   build_model:
     cmd: python build_model.py
@@ -27,8 +27,8 @@ stages:
       size: 5134
     - path: data/prepared_data
       hash: md5
-      md5: 3767eec56906f5ac724a3f07433645ef.dir
-      size: 13442342
+      md5: ec064b0274e2e6a0864580a748e6bb6a.dir
+      size: 24067192
       nfiles: 2
     params:
       configs/build_model.yaml:
@@ -36,8 +36,8 @@ stages:
           output_filepath: ./data/model/autogluonmodel/
           problem_type: regression
           eval_metric: mean_absolute_error
-          time_limit: 400
-          presets: good_quality
+          time_limit: 60
+          presets: medium_quality
           excluded_model_types:
           - KNN
         SKLearnLinearRegression:
@@ -49,25 +49,25 @@ stages:
     outs:
     - path: data/model/
       hash: md5
-      md5: 7b2f8334c81fb5ff23e42e77741b31d1.dir
-      size: 118227750
-      nfiles: 71
+      md5: dc73587056f07735719bfee464a5f898.dir
+      size: 285397707
+      nfiles: 16
     - path: metrics/fit_metrics.json
       hash: md5
-      md5: e1c9a16617804f48e8ffac7cec6575ca
-      size: 185
+      md5: f6d03cb197a3d78e61f6fef023ed8d7f
+      size: 184
   generate_predictions:
     cmd: python generate_predictions.py
     deps:
     - path: data/model
       hash: md5
-      md5: 7b2f8334c81fb5ff23e42e77741b31d1.dir
-      size: 118227750
-      nfiles: 71
+      md5: dc73587056f07735719bfee464a5f898.dir
+      size: 285397707
+      nfiles: 16
     - path: data/prepared_data
       hash: md5
-      md5: 3767eec56906f5ac724a3f07433645ef.dir
-      size: 13442342
+      md5: ec064b0274e2e6a0864580a748e6bb6a.dir
+      size: 24067192
       nfiles: 2
     - path: generate_predictions.py
       hash: md5
@@ -83,21 +83,21 @@ stages:
     outs:
     - path: data/predictions/
       hash: md5
-      md5: fb7cf3f4a90598ec1e43a1b7a4af3bef.dir
-      size: 536774
+      md5: 3d70862c00e38f383a06c3e4df5ebb38.dir
+      size: 386958
       nfiles: 1
   generate_metrics:
     cmd: python generate_metrics.py
     deps:
     - path: data/predictions
       hash: md5
-      md5: fb7cf3f4a90598ec1e43a1b7a4af3bef.dir
-      size: 536774
+      md5: 3d70862c00e38f383a06c3e4df5ebb38.dir
+      size: 386958
       nfiles: 1
     - path: data/prepared_data
       hash: md5
-      md5: 3767eec56906f5ac724a3f07433645ef.dir
-      size: 13442342
+      md5: ec064b0274e2e6a0864580a748e6bb6a.dir
+      size: 24067192
       nfiles: 2
     - path: generate_metrics.py
       hash: md5
@@ -106,14 +106,12 @@ stages:
     params:
       configs/generate_metrics.yaml:
         dataclient_type: local
-        input_datahandler_type: parquet
         metrics_output_filepath: ./metrics/metrics.json
         metrics_type: Regression
-        output_datahandler_type: json
     outs:
     - path: metrics/metrics.json
       hash: md5
-      md5: 852ef4cf2ca5e7f89d70420a9df7a596
+      md5: 2507f756ea68768185ebeaf66db2ebbd
       size: 183
   startup_cleanup:
     cmd: python startup_cleanup.py

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

@@ -5,5 +5,3 @@ autogluon==0.8.2
 alibi==0.9.4
 pyarrow==13.0.0
 pre-commit==3.3.3
-sphinx==7.2.5
-sphinx_rtd_theme==1.3.0