Churn Analysis with Machine Learning Classification Algorithms in Python

Abstract

“Artificial Intelligence” is the ability of a computer or a computer controlled robot to perform various activities in a similar way within intelligent creatures. The word “AI”, which is the acronym of artificial intelligence concept, is also frequently used in informatics. “Artificial Intelligence” studies are generally aimed at developing similar artificial instructions by analyzing human thinking methods. Also, from a point of view, even though it may seem like an attempt by a programmed computer to think, these definitions are changing rapidly, and new orientations are formed towards the concept of “Artificial Intelligence” that can learn and develop independently from human intelligence in the future. Also, “Machine Learning” is a method paradigm that makes inferences from existing data using mathematical and statistical methods and makes predictions about the unknown with these inferences, such as face recognition, document classification, spam detection and etc. In addition, nowadays, many big companies such as Google, Facebook and especially Tesla continue their research on “Artificial Intelligence” and “Machine Learning” under the title of “Churn Analysis”. The purpose of “Churn Analysis” is to identify the customers who are likely to stop using their service or product. Today’s competitive conditions have formed many companies that try to sell the same product very close to each other. Therefore, the existing customer is very valuable. Customer loss can be a significant loss of revenue, especially in environments where customer preferences do not change very often, such as financial services. Moreover, with the help of “Churn Analysis”, customers’ escaping probability with precise estimates can be predicted by saying that the customers’ leaving probabilities are. Also, some analysis can be made based on customer segments and loss amount – money’s worth. After these analyses, communication can be improved and engagement can be increased to keep the customers. Furthermore, effective marketing campaigns can be created in order to target the customers by calculating the rate of churn or wear. This can significantly increase profitability of the companies. In this paper, the terms “Artificial Intelligence” and “Machine Learning” in computer science and the term “Churn Analysis” in global marketing have been tried to be combined together. For this purpose, 7043 (11 missing data – 7032 in total) user data with 21 different features – customerID, Gender, SeniorCitizen, Partner, Dependents, Tenure, PhoneService, MultipleLines, InternetService, OnlineSecurity, OnlineBackup, DeviceProtection, TechSupport, StreamingTV, StreamingMovies, Contract, PaperlessBilling, PaymentMethod, MonthlyCharges, TotalCharges and Churn – belonging to Telco Avi (in Telecom Company) have been retrieved and extracted from “Kaggle Data Science”. Finally, this user data set has been processed, treated and committed by k-Nearest Neighbors (KNN) Algorithm, Artificial Neural Networks (ANN) Algorithm, Gaussian Naive Bayes Algorithm and Random Forests Algorithm, which are the “Machine Learning Classification Algorithms” in “Artificial Intelligence” – in Python by using its own libraries – Pandas, NumPy, Keras, Sklearn, Matplotlib, Seaborn, etc. – in order to recognize and learn the “Churn Analysis” in the company – Telco Avi.