What is the K Nearest Neighbors?

It is an algorithm that can be used for both classification and regression. It is named "nearest neighbors" because the principle of this model is indeed to choose the k data closest to the point studied in order to predict its value.

How we use it for our problems?

Thanks to python and the data preprocessing done before, the implementation will be quite fast. We are going to use the famous python library for the data science, Scikit-Learn. It has many machine learning algorithm including the K Nearest Neighbors named KNeighborsClassifierin the library.

Firstly we need to define a target, i.e the feature that we want to predict. Here it will be obviously the churn. Then we will divide our dataset in 2 parts : 80% will serve to train the model and 20% to test the model. After each test we can calculate an accurate score which indicate the precision of our model. Finally we will try many values of k to figure out which one will be the best for our model.

What is a Decision Tree?

It's a method based on the use of a decision tree as a predictive model. In these tree structures, the leaves represent the values of the target variable and the branches correspond to combinations of input variables that lead to these values.

How we use it for our problems?

We are going to use the same schema than for K Nearest Neighbors to implement our algorithm. Except that this time we don't have to create our target (Already create before) and that we will try it only one time because there isn't parameter to change unlike previously.

This time our model made no mistake in his predictions as the accuracy score shows(100%). But we have to stay prudent with this result. The perfect model doesn't exist and as we stated for the association rules, the dataset need to have more data with churn to be more efficient.