Machine Learning-Customer Churn Prediction
- Project Name : Machine Learning-Customer Churn Prediction.
- Category : Data Analysis
- Complete Date : 15 September, 2021
- Skills : Pthon, random-forest, pca, classification logistic-regression. decision-trees, class-imbalance, svm-classifier, telecom-churn-prediction, telecom-churn-analysis
Machine Learning-Customer Churn Prediction
A well-known telco has been observing a lot of customers not using their mobile numbers or switching to competitor telcos over the past couple of quarters. This has caused a huge dent in their quarterly revenues and might drastically affect annual revenues for the ongoing financial year, causing stocks to plunge and market cap to reduce significantly. The idea is to be able to predict which customers are going to churn so that necessary actions/interventions can be taken by the telco to retain such customers.
Project Description
In this machine learning churn prediction project, we are provided with customer data pertaining to his past call history with the telco and some demographic information. We use this to establish relations/associations between data features and customer’s propensity to churn and build a classification model to predict whether the customer will leave the telecom service or not. We also go about explaining model predictions through multiple visualizations and give insight into which factor(s) are responsible for the churn of the customers.
Project Challenge
This project walks you through a complete end-to-end cycle of a data science project in the telco industry, right from the deliberations during formation of the problem statement to making the model deployment-ready.
- Demographic information has anomaly.
- No significant issues were predicted.
- Competitor were complaining same that where they go.
- Churn were not so huge (5%-10(%)
Project Solution
since the rate of churn is typically low (about 5-10%, this is called class-imbalance) – we were using techniques to handle class imbalance. We took the following suggestive steps to build the model:
- Preprocess data (convert columns to appropriate formats, handle missing values, etc.)
- Conduct appropriate exploratory analysis to extract useful insights (whether directly useful for business or for eventual modelling/feature engineering).
- Derive new features.
- Reduce the number of variables using PCA.
- Train a variety of models, tune model hyper parameters, etc. (handle class imbalance using appropriate techniques).
- Evaluate the models using appropriate evaluation metrics. Note that it is more important to identify churners than the non-churners accurately – choose an appropriate evaluation metric which reflects this business goal.
- Finally, choose a model based on some evaluation metric.
We will be in great pleasure to have opportunity to be a part of your growth