254 add CLI functions for ML models

[nmaarnio]

Added:

• 7 CLI functions:

  • random_forest_classifier_train_cli
  • random_forest_regressor_train_cli
  • gradient_boosting_classifier_train_cli
  • gradient_boosting_regressor_train_cli
  • logistic_regression_train_cli
  • evaluate_trained_model_cli
  • predict_with_trained_model_cli

• Notebook called machine_learning_rf_gb_lr.ipynb. The data used in the notebook is not available in remote test data folder, but the training data is an area within the IOCG test data region, and the test data is the whole IOCG test data. Clipping, unifying, rasterization and computing distances to structures were done as preprocessing. NOTE: The logistic regression model did not predict any deposits in the area – if someone can comment / figure out the reason, that would be great. There might be (other) mistakes in the notebook too as I am not very familiar with ML modeling.

• A module called machine_learning_general.py. I combined here all ML utilities that were previously in different files. New additions are prepare_data_for_ml and reshape_predictions utility functions which aim to streamline performing ML modeling with EIS Toolkit

[RichardScottOZ]

On the logistic regression comment above - why would you think a basic model without much done to it to produce interesting results? I wouldn't think a simple thing with a few AEM features would do much with a model of this type, in gneral.

Added:

• 7 CLI functions:

  • random_forest_classifier_train_cli
  • random_forest_regressor_train_cli
  • gradient_boosting_classifier_train_cli
  • gradient_boosting_regressor_train_cli
  • logistic_regression_train_cli
  • evaluate_trained_model_cli
  • predict_with_trained_model_cli
• Notebook called machine_learning_rf_gb_lr.ipynb. The data used in the notebook is not available in remote test data folder, but the training data is an area within the IOCG test data region, and the test data is the whole IOCG test data. Clipping, unifying, rasterization and computing distances to structures were done as preprocessing. NOTE: The logistic regression model did not predict any deposits in the area – if someone can comment / figure out the reason, that would be great. There might be (other) mistakes in the notebook too as I am not very familiar with ML modeling.
• A module called machine_learning_general.py. I combined here all ML utilities that were previously in different files. New additions are prepare_data_for_ml and reshape_predictions utility functions which aim to streamline performing ML modeling with EIS Toolkit

On the logistic regression comment above - why would you think a basic model without much done to it to produce interesting results? I wouldn't think a simple thing with a few AEM features would do much with a model of this type, in general.