Classification Project

Project Description

Companies all across the globe deal with customers leaving on a regular basis. These people are described as those who "churn". Telco is wondering, "Why are our customers leaving?" In this project I will look at data of their customers and identify drivers that might affect the chance a customer might churn.

Project Goals

• Identify drivers of churn for Telco customers
• Use the identified drivers to develop a model to determine if a customer is likely to churn
• Churn is defined as a customer who has left Telco
• This information can be useful to help Telco find ways to retain customers who are likely to churn.

Initial Thoughts

My initial hypothesis is that drivers of churn will be predictors of if a customer will churn in the near future.

The Plan

• Aqcuire the data from Codeup mySQL database

• Prepare data

  • Remove redundant data columns (payment_type_id', 'internet_service_type_id', 'contract_type_id', 'customer_id', )
  • Drop customers whose total_charges are spaces; these customers are new customers (tenure=0 and churn=No)
  • Encode churn as 'churn_encoded' {'Yes': 1, 'No': 0}, then drop 'churn' column

• Explore data in search of drivers of churn

  • Answer the following initial question
    • How often do customers churn?
    • Does a customer's tenure affect churn?
    • Do higher monthly charges affect churn?
    • Does contract type affect churn?
    • Do those with no tech support churn more?

• Develop a model to predict if a customer will churn

  • Use drivers identified through exploration to build different predictive models
  • Evaluate models on train and validate data
  • Select best model base on highest recall (because saying they didn't churn and they churned is more costly)
  • Evaluate the best model on the test data

• Draw conclusions

Data dictionary

Feature	Definition	Values
gender	The customer's gender	Male or Female
senior_citizen	Is this customer a senior citizen?	{'Yes': 1, 'No': 0}
partner	Does this person live with a partner?	'Yes', 'No'
dependents	Does this person live with dependents?	'Yes', 'No'
tenure	The number of months a customer has been with the company	float
phone_service	Does this person subscribe to phone service?	'Yes', 'No'
multiple_lines	Does this person have multiple phone lines?	'Yes', 'No', 'No phone service'
online_security	Does this person subscribe to online security?	'Yes', 'No', 'No internet service'
online_backup	Does this person subscribe to online backup?	'Yes', 'No', 'No internet service'
device_protection	Does this person subscribe to device protection?	'Yes', 'No', 'No internet service'
tech_support	Does this person subscribe to tech support?	'Yes', 'No', 'No internet service'
streaming_tv	Does this person subscribe to streaming TV?	'Yes', 'No', 'No internet service'
streaming_movies	Does this person subscribe to streaming movies?	'Yes', 'No', 'No internet service'
paperless_billing	Does this person use paperless billing?	'Yes', 'No'
monthly_charges	The amount a customer is currently charged per month	float
total_charges	The amount a customer has been charged since becoming a customer	float
contract_type	The length of contract the customer currently has	'Month-to-month', 'One-year', or 'Two-year'
internet_service_type	Type of internet service	'DSL', 'Fiber', 'None'
payment_type	The way the customer pays their bill.	'Mailed check', 'Electronic check', 'Credit card (automatic)', 'Bank transfer (automatic)'
Target variable
churn_encoded	Did the customer leave the company?	{'Yes': 1, 'No': 0}

Steps to Reproduce

1. Clone this repo
2. Acquire the data from Codeup mySQL "telco" database using your personal env.py file where you store your username, password, and host
3. Put the data in the file containing the cloned repo.
4. Run notebook.

Takeaways and Conclusions

• About 1/4 of our customers churn
• Tenure seems to be a driver of churn; we need to intice customers to stay around.
• Monthly_charges seem to be a driver of churn; we need to help lower monthly charges.
• Contract_type seems to be a driver of churn; we need to intice customers to sign longer contracts.
• Tech_support seems to be a driver of churn; we should encourage more customers to get tech support

Recommendations

• To reduce churn, I would encourage the company to think about taking the following actions:
  • Seek out young customers and give discounts (lower monthly cost) in order to keep customers longer (increase tenure)
  • Seek out month-to-month contract customers and encourage them to sign a longer contract maybe through discounts
  • Seek out those who don't have tech support and give a free trial of tech support.

Next Steps

• In the next iteration:
  • Look into how customers with internet add-ons are the same people then can see if they churn more or less.
  • Look into cost of add-ons and see if there are bundles that might reduce monthly charges. That could help us entice more customers to stay.