Business Problem:

Overfitting is a common problem with machine learning models especially when we have just a few training datapoints. The lesser the number of train data points, the less able is our model to generalize on the unseen or the test data points.

Hence, we need to be careful while the training process and see how our model performs. By just getting an accuracy of about 90% on train data we cannot assume that our model will perform the same on unseen dataset.

This is the problem we are going to deal with today. We will also see by how using the simple machine learning models like KNeighborsClassifier and LogisticRegression we can reduce overfitting and help our model generalize better on unseen data even with a less amount of training data that we have.

#feature-engineering #sklearn #machine-learning #overfitting #artificial-intelligence