Quick add use of pipeline in performing standard scaling and then decision tree to fit the given dataset

Comment out experiment code

Add in RandomizedSearchCV step by the end

Adding in visualization for each class examining the cross relationships of variables

Adding in a step of performing StandardScaling in the presentation

Upload written out Jupyter Notebook markdown file that has written out points regarding the class project

Perform Fine Tuning Step 1. Perform Cross-Validation to evaluate performance 2. Using RandomizedSearchCV to perform systematic tuning. This identify the best parameter that is applicable for the model chosen

1. Minor correction with spacing 2. Add in section regarding how to perform cross-validation to evaluate performance 3. Picked out particular metrics interested for multi-class classification to examine

1. Clean up variable for clarity 2. Re train SVC model with the probability parameter set to True for micro-average and macro-average calculation 3. Calculate micro-average and macro-average metrics for SVC models built and graph out the ROC

Clean up import lines and reorgnize where the import are performed

1. Define common function to perform same exact micro-averaged and macro-averaged metric calculation to be used across models 2. Define common plot function to create the ROC Curve plotting across multiple models 3. Write driver code to iterate through each model built for the 1. and 2. defined earlier

1. Examine macro vs micro metrics using metrics module AUC and ROC_Curve class 2. Create the ROC Curve using RocCurveDisplay class applying the one-vs-rest approach since dealing with multi-class classification

Add code to examine the class imbalance, confusion matrix using the package sklearn-evaluation. Note that installation should only be performed once and have a tag 'do_not_execute'

Add code to compare the result with actual testing dataset class using a classification report. Do this for each one of the model built

1. Investigation into why the length of vector is different from prior models and having additional entry 2. Examine the prediction output in 2 different ways one being exclude the predicted category from evaluation, aka reducing number of rows or second way being include the predicted category from evaluation, aka extendng the existing y_test with transformation to include full class set