Cayt-Schlichting/example-classification-project

Project Summary

Project Goal: The goal of this project is to identify key drivers of customer churn in order to make recommendations on how to minimize future churn. We aim to create a model to identify customers at a higher risk of churn and use this to take preventative action.

• Identify key drivers of churn
• Develop and test classification models for identifying customers at risk of churn

Initial Hypothesis:

• I expect churn to be primarily affected by:
  • Cost: Higher cost customers will be more likely to leave.
  • Contract: Customers are less likely to leave if they have a 1 or 2 year contract.
  • Alternatives: Internet customers without alternatives will be less likely to leave.
    • NOTE: This dataset does not have the information to test this hypothesis, however the availability of alternatives is an important unknown to keep in mind.

Initial Questions:

• What features are the key drivers of churn?
• Which of these features can be used to predict churn?

Project Plan:

• Acquire data from servers
• Prepare Data:
  • Identify any duplicate or unusable columns
  • Identify any missing data
  • Identify any needed transformation
  • Takes steps to clean data
• Combine all data acquisition and preparation steps into modules
• Explore data
  • subset non-encoded data for visualizations
  • create visual representations of all features in relation to target variable
  • perform hypothesis testing on subset of features that visually appear to be drivers of churn
  • conduct some multivariate analysis, primarily using features identified as drivers of churn
• Model
  • subset encoded data for modeling
  • Create baseline
  • Determine model methods to use and hyperparameters for each. Minimum of 10 models.
  • Generate and fit these models on the train dataset
  • Run model statistics to identify top performing models (3-5)
  • Run this subset of modoels against the validate subset
  • Evaluate which model performed the best based off preferred model statistics and performance across subsets
• Generate Final Report
  • Include subset of details across project steps
  • run model over test dataset

Deliverables:

• wrangle.py module with preparation and acquisition functions
  • wrangle_notebook.ipynb contains steps and notes on data acquisition and preparation decisions
• eda.ipynb
  • Contains exploratory analysis of the data including visualizations and hypothesis testing
• modeling.ipynb
  • Contains full modeling work, with notes on parameter choices and model evaluation
• Final_Report.ipynb
  • Contains curtailed version of project in presentable format.
• predictions.csv
  • Contains best model's predictions and probability on test subset

Reproducing the project:

• User will need an env.py file with the the following variables: 'host', 'user' (username), and 'password'
• Data preparation assumes explicit list of columns and database tables. Changes to the underlying database may require updated data cleaning
• The additional files contain

Data Dictionary:

This does not contain a list of the categorical columns after encoding

Column	Non-Null Count	Dtype	Description and Values
gender	7043 non-null	object	gender: ['Female' 'Male']
senior_citizen	7043 non-null	int64	[0 1]
partner	7043 non-null	object	['Yes' 'No']
dependents	7043 non-null	object	['Yes' 'No']
tenure	7043 non-null	int64	Months customer has been with Telco
phone_service	7043 non-null	object	['Yes' 'No']
multiple_lines	7043 non-null	object	If the customer has multiple phone lines: ['No' 'Yes' 'No phone service']
online_security	7043 non-null	object	['Yes' 'No' 'No internet service']
online_backup	7043 non-null	object	['Yes' 'No' 'No internet service']
device_protection	7043 non-null	object	['Yes' 'No' 'No internet service']
tech_support	7043 non-null	object	['Yes' 'No' 'No internet service']
streaming_tv	7043 non-null	object	['Yes' 'No' 'No internet service']
streaming_movies	7043 non-null	object	['Yes' 'No' 'No internet service']
paperless_billing	7043 non-null	object	Whether the customer uses paperless billing: ['Yes' 'No']
monthly_charges	7043 non-null	float64	Monthly charge of customer (dollars)
total_charges	7043 non-null	object	Total charges in customer lifetime (dollars)
churn	7043 non-null	object	Whether the customer churned last month: ['Yes' 'No']
internet_service_type	7043 non-null	object	Internet service customer uses: ['DSL' 'Fiber optic' 'None']
contract_type	7043 non-null	object	Contract of customer: ['One year' 'Month-to-month' 'Two year']
payment_type	7043 non-null	object	Payment method customer uses: ['Mailed check' 'Electronic check' 'Credit card (automatic)' 'Bank transfer (automatic)']