/sklearn_wrapper

A wrapper to simplify the ussage of sklearn for massive experimentation with different modeling methods

Primary LanguagePythonMIT LicenseMIT

sklearn_wrapper

A wrapper to simplify the usage of sklearn for massive experimentation with different modeling methods that provides easy access to the model parameters

Currently one class is supported for classification methods

Modeling Class

Init Method

  • You should pass the train data and labels when initializing. The you can run all modeling methods on these data
  • Optionally you can also pass the test data and labels so they are alighed and always be tested together

MOdeling Methods

  • SVM
  • Random Forest
  • Decision Tree
  • Gradient Boosting
  • Linear Regression
  • Elastic Net Regression
  • HMM with GMM Probabilities - requires hmmlearn lib

Other Methods

  • Save Model
    • Stores model name
    • Model Hyperparameters
    • The actual Model
    • The performance results if the model has been tested and results have been added into model.results dict - examples:
    model.results['accuracy'] = accuracy_value
model.results['r^2'] = rSquared_value
model.results['custom_metric_name'] = custom_metric_value

    • Will update soon to optionally store the train and test data and labels

More methods and functionalities will be added gradually in the near future