Telco Project

Project Description

• This project uses the telco data set from MySql and is used for practice on figuring out why customers are churning using the data science pipeline.

Project Goal

• Discover which customers are more likely to churn based of the different features Use machine learning model to classify customers likely to churn.

Initial Thoughts

Need to figure out what features will lead me to the most churn based off the customers who have churned and create/fit a model that will find probable churn.

The Plan

• Acquire data from sql

• Prepare Data

  • Drop unnecessary columns(Update list here)
  • Clean up the number values
  • Encode values for the features used in the future

• Explore

  • What features did the customers who churned had or didn't have
  • What features do long term customers have
  • List of Potential churn items From a simple barchart

• Modeling

  • Use drivers in explore to build predictive models of different types
  • Evaluate models on train and validate data
  • Select the berst model based on accuracy
  • Evaluate the test data

Data Dictionary

| Feature | Definition |
|:--------|:-----------|
|Customer ID| Customer's unique identifier|
|Gender| Customer's gender|
|Senior_Citizen| Whether the customer is a senior citizen or not|
|Partner| Whether the customer has a partner or not|
|Dependents| Whether the customer has dependents or not|
|Tenure| Number of months customer has been with company in whole numbers|
|Online Security| OnlineSecurity: Whether the customer has online security or not|
|Online Backup| Whether the customer has online backup or not|
|Device Protection| Whether the customer has device protection or not|
|Tech Support| Whether the client has tech support or not|
|Churn| Has the customer churned|
|Additional Features| Encoded and values for categorical data and scaled versions continuous data|

How to Reproduce

• Clone this repo
• Acquire data from MySql (It should make a telco.csv after)
• Run Notebook

Takeaways and Conclusions

• Most churn: No Online Security, No Online Backup, No Device Protection, No Tech Support, Month-to-Month Contract Types, Fiber Optic Internet Service Type, Electronic Check Payment Types

• No significant difference in churn(per visuals): Gender, Partner, Phone Service, Multi Lines, Streaming TV or Movies, Paperless Billing, Auto Payments and Mailed Checks.

• Logistic Regression model performed the best on the accuracy for determining churn

Recommendations

• I would recommend that somehow we push customers towards getting Online Security and Backup, Tech Support, and Device Protection.

• I would also recommend that customers are driven towards DSL and automatic Payments like Bank Transfer or Credit Card.