We have seen how decision trees work. Let's try to solve a classification problem using decision trees.

Write a function load_data that loads a dataset with numpy's loadtext api

• Accepts the following parameter
  • path to file (str)
  • skiprows (header rows to be skipped)

• Accepts the following parameters:

  • X_train, y_train, X_test (Numpy arrays for training, testing; any format acceptable by sklearn will work)
  • paramgrid (list of parameters (including those of the classfier) for RandomizedSearchCV)
  • n_iter_search (Number of iterations the search will be run)
  • KFold (the number of k-folds to be used in cross-validation) (Optional) (Default 3)
  • early_stopping_rounds (Int) (Optional) (Default 10)
  • seed (a number; a subsequent call to the function with the same seed will reproduce the same results)(optional) (Default 42)
  • **kwargs (To set parameters to the base classifier)

• Should return

  • predictions for X_test
  • trained RandomizedSearchCV object

• Write a function called finetune_class which

  • Takes in X_train, X_test, y_train, param_grid, n_iter_search
  • Returns y_pred_test, Trained RandomizedSearch Object

• You will use myDecisionClassifier() function

• Based on the stage-wise optimization values further fine tune the model

• You will need to provide a parameter grid to the said function, be careful to choose which values to be optimized.