Evaluating Logistic Regression

Let's now evaluate the logistic regression model that we built previously.

Exercise 70: Evaluate a Logistic Regression Model

Machine learning models fitted on a training dataset cannot be evaluated using the same dataset. We would need to leverage a separate test dataset and compare the model's performance on a train as well as a test dataset. The caret package has some handy functions to compute the model evaluation metrics previously discussed.

Perform the following steps to evaluate the logistic regression model we built in Exercise 7, Build a Logistic Regression Model:

1. Compute the distribution of records for the RainTomorrow target variable in the df_new DataFrame:

   print("Distribution of labels in the data-")
   print(table(df_new$RainTomorrow)/dim(df_new)[1])

   The output is as follows:

   "Distribution of labels in the data-"
          No       Yes
   0.7784459 0.2215541

2. Predict the RainTomorrow target variable on the train data using the predict function and cast observations...