FINDINGS AND DISCUSSION
The experiment was carried out in two phases:
In this first phase, the data was modeled using the GLM model based on
the Binomial random component, which specified the use of the Binomial
probability distribution function for the response variable. The link
function used here was the Logit function, which is part of the logistic
regression to model categorical variables because, in the experiment, we
are to determine whether, based on the available variables, it will rain
or not. The training accuracy for the model was 84%, while the testing
accuracy was 83%. The experiment results are as follows