Skip to content
/ Churn-Prediction-Model Public

Churn Prediction Model with EDA, Feature Engineering and testing out different models. Based on a kaggle notebook.

2 stars 0 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

raj-pulapakura/Churn-Prediction-Model

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
README.md
README.md
 
 
Trial 1_ Churn Prediction EDA.ipynb
Trial 1_ Churn Prediction EDA.ipynb
 
 
Trial 1_ Churn Prediction Model 1 (base model).ipynb
Trial 1_ Churn Prediction Model 1 (base model).ipynb
 
 
Trial 1_ Churn Prediction Model 2 (balancing dataset).ipynb
Trial 1_ Churn Prediction Model 2 (balancing dataset).ipynb
 
 
Trial 1_ Churn Prediction Model 3 (modifying network architecture).ipynb
Trial 1_ Churn Prediction Model 3 (modifying network architecture).ipynb
 
 
Trial 1_ Churn Prediction Model 4 (XGBoost).ipynb
Trial 1_ Churn Prediction Model 4 (XGBoost).ipynb
 
 
Trial 1_ model1.json
Trial 1_ model1.json
 
 
Trial 1_ model1_weights(1).hdf5
Trial 1_ model1_weights(1).hdf5
 
 
Trial 1_ model1_weights.hdf5
Trial 1_ model1_weights.hdf5
 
 
Trial 1_ model2.json
Trial 1_ model2.json
 
 
Trial 1_ model2_weights.hdf5
Trial 1_ model2_weights.hdf5
 
 
Trial 1_ model3.json
Trial 1_ model3.json
 
 
Trial 1_ model3_weights.hdf5
Trial 1_ model3_weights.hdf5
 
 
Trial 1_ output_cleaned_dataset.csv
Trial 1_ output_cleaned_dataset.csv
 
 
Trial 2_ Churn Prediction Analysis.ipynb
Trial 2_ Churn Prediction Analysis.ipynb
 
 
Trial 2_ Churn Prediction Model 1 (base model).ipynb
Trial 2_ Churn Prediction Model 1 (base model).ipynb
 
 
Trial 2_ Churn Prediction Model 2 (balancing dataset).ipynb
Trial 2_ Churn Prediction Model 2 (balancing dataset).ipynb
 
 
Trial 2_ Churn Prediction Model 3 (modifying network architecture).ipynb
Trial 2_ Churn Prediction Model 3 (modifying network architecture).ipynb
 
 
Trial 2_ Churn Prediction Model 4 (XGBoost).ipynb
Trial 2_ Churn Prediction Model 4 (XGBoost).ipynb
 
 
Trial 2_ model1.json
Trial 2_ model1.json
 
 
Trial 2_ model1_weights.hdf5
Trial 2_ model1_weights.hdf5
 
 
Trial 2_ model2.json
Trial 2_ model2.json
 
 
Trial 2_ model2_weights.hdf5
Trial 2_ model2_weights.hdf5
 
 
Trial 2_ model3.json
Trial 2_ model3.json
 
 
Trial 2_ model3_weights.hdf5
Trial 2_ model3_weights.hdf5
 
 
Trial 2_ output_cleaned_dataset.csv
Trial 2_ output_cleaned_dataset.csv
 
 
telecom_customer_churn.csv
telecom_customer_churn.csv
 
 
telecom_data_dictionary.csv
telecom_data_dictionary.csv
 
 
telecom_zipcode_population.csv
telecom_zipcode_population.csv
 
 

Repository files navigation

Churn-Prediction-Model

Churn Prediction Model with EDA, Feature Engineering and testing out different models.

Test it out: https://www.kaggle.com/code/rajpulapakura/churn-prediction-model/

What is churn prediction?

Churn prediction is the process by which a business attempts to predict which customers are at high risk of leaving the company or cancelling a subscription to a service, based on their behaviour with the business' product.

Why is it important?

  • The cost of acquiring a new customer is almost always greater than the cost to retain an existing customer.
  • This implies that businesses should prioritise retaining its existing customers.
  • Customer churn informs a business if a customer is likely to leave the company.
  • The business can then take further action to retain the customer such as targeted re-engagement campaigns or even re-evaluating company-wide retention initiatives like pricing.
  • Predicting churn allows for the identification of factors that lead to churn.
  • Predicting churn helps in preventing churn

What is this project about?

This project takes a dataset from kaggle:

https://www.kaggle.com/datasets/shilongzhuang/telecom-customer-churn-by-maven-analytics

This dataset contains 2 tables, in CSV format:

  • The Customer Churn table contains information on all 7,043 customers from a Telecommunications company in California in Q2 2022
  • Each record represents one customer, and contains details about their demographics, location, tenure, subscription services, status for the quarter (joined, stayed, or churned), and more!
  • The Zip Code Population table contains complimentary information on the estimated populations for the California zip codes in the Customer Churn table

The aim of the project is to create a churn prediction model to predict which customers are likely to leave the company. This is done through classification (0 = Customer will stay, 1 = Customer will churn).

Processes used:

  • EDA: In depth analysis of the dataset features
  • Feature Engineering: Creating new features which aim to accurately model the relationship between the original features and the target variable
  • Testing out different models: Several unique models were utilised throughout the project - RandomForestClassifier, Neural Networks and XGBoost

What is included in the project?

The project includes the files used for data analysis and model testing. I conducted two trial runs of the entire process (EDA -> Model Testing).

How can you test it out?

I have summarised the best results and data analysis in a kaggle notebook which you can find by clicking this link:

https://www.kaggle.com/code/rajpulapakura/churn-prediction-model/

About

Churn Prediction Model with EDA, Feature Engineering and testing out different models. Based on a kaggle notebook.

Topics

machine-learning ai deep-learning neural-network eda kaggle xgboost classification data-processing churn-prediction

Resources

Readme
Activity

Stars

2 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages