Good day and welcome to half three of Venture Three. On this mission, we’re specializing in classification. The aim is to create a mannequin that may predict whether or not a constructing is severely broken or not severely broken in an earthquake. Within the earlier half, we realized about logistic regression. Immediately, we shall be studying about one other kind of mannequin: the choice tree.
Our three targets for immediately are:
1. Create a call tree mannequin to foretell extreme injury.
2. Tune the hyperparameters for that mannequin, specializing in one explicit hyperparameter.
3. Clarify our mannequin predictions utilizing one thing known as Gini significance.
Following the ML workflow, for importing, we’ll hold every little thing the identical and use the identical wrangle perform. We cannot conduct extra EDA. As an alternative, we’ll carry out a three-way cut up of our information into practice, validation, and take a look at units beneath splitting.
We’ll proceed to work on the accuracy rating as our baseline for constructing the mannequin. Within the iteration part, we’ll cowl three most important matters: determination timber, ordinal encoding for categorical variables, and plotting a validation curve to assist tune the hyperparameters for our determination tree mannequin.