Overview run_ml_job is Stage 3a of the pipeline (alternative to run_dl_job). It triggers the ml-worker-job pipeline job, which trains a classical machine learning model using scikit-learn on the feature-engineered dataset and generates per-ticker price predictions.The tool blocks until the job completes and returns the output blob URL.
Parameters Blob storage URL pointing to a feature_engine_*.json file. This is the output_url returned by run_feature_worker.
Blob storage URL pointing to a data_extractor_*.json file. This is the output_url returned by run_data_extraction. Required alongside feature_url
for proper train/test date alignment.
ML model configuration. Model selection and cross-validation settings. Model type. Accepted values (case-insensitive, spaces/dashes normalized): Value Algorithm "random_forest" / "randomforest" / "rf"Random Forest "xgboost" / "xgb"XGBoost "lightgbm" / "lgbm"LightGBM "catboost"CatBoost "linear" / "linear_regression"Linear Regression "ridge"Ridge Regression "lasso"Lasso Regression "elastic_net" / "elasticnet"Elastic Net "quantile"Quantile Regression
Unknown values fall back to "random_forest". Number of estimators for tree-based models (Random Forest, XGBoost, LightGBM, CatBoost).
Default: 100.
Number of samples in the test fold for cross-validation.
Pass an integer (e.g. 15 = 15 days). Default: 15. If you pass a float less than 1 (e.g. 0.2), it is automatically converted to the default 15.
Always use an integer.
Number of cross-validation splits (time-series walk-forward). Default: 15.
Returns { "status" : "Succeeded" , "output_url" : "https://stmcpfabricdev.blob.core.windows.net/data/ml_engine_20260307_125841.json" , "output_name" : "ml_engine_20260307_125841.json" , "execution_name" : "ml-worker-job-abc123xyz" }
Field Description statusJob terminal status (Succeeded) output_urlFull HTTPS URL to the output blob — pass to run_po_job output_nameBlob filename execution_nameJob execution ID for audit/debugging
Example { "feature_url" : "https://stmcpfabricdev.blob.core.windows.net/data/feature_engine_20260307_125103.json" , "data_extractor_url" : "https://stmcpfabricdev.blob.core.windows.net/data/data_extractor_20260307_124937.json" , "config" : { "ML params" : { "model" : "random_forest" , "n_estimators" : 100 , "test_size" : 15 , "n_splits" : 15 } } }
Resources Resource Value Container Apps Job ml-worker-jobContainer name ml-worker-jobEnv vars injected FEATURE_URL, DATA_EXTRACTOR_URL, CONFIGOutput blob prefix ml_engine_Timeout 600 seconds
Next Step Pass output_url to run_po_jobinput_url.
If you prefer a neural network approach, use run_dl_jobnn_engine_*.json blob that is equally compatible with run_po_job. 
