RachelZoe27 / starbucks-statistical-promotion-analysis

A/B test on promotion effect & optimization model for future promotion strategy using data from Starbucks.

Starbucks Statistical Promotion Analysis & Optimization

1. Installation
2. Objective
3. File Descriptions
4. Licensing, Authors, Acknowledgements

Installation

The code was developed using Python 3.6.3. Necessary packages beyond the Python Standard Library are:

• numpy==1.12.1
• pandas==0.23.3
• scipy==1.2.1
• scikit-learn==0.19.1
• matplotlib==2.1.0

Objective

In this project, an A/B test on customer promotion is analyzed. Predefined KPIs are statistically evaluated.

KPIs:

• Incremental Response Rate (IRR)
• Net Incremental Revenue (NIR)

After exploratory and statistical analysis, a classification model based on customers' features is fitted to the data in order to optimize future promotion strategy. The goal is to target the most promising customers regarding IRR & NIR.

More details on the questions of interest can be taken from the Jupyter Notebook introduction.

File Descriptions

Starbucks.ipynb

• notebook containing a detailed description of the questions of interest
• statistical analysis & optimization model

training.csv

• Starbucks training dataset used in the notebook (the dataset has been used for assessments of data scientist job interviews at Starbucks in the past)

test_results.py

• test on optimization model provided on Udacity.

Test.csv

• Starbucks test dataset used in test_results.py

Licensing, Authors, Acknowledgements

The data used in this project was kindly provided by Starbucks in cooperation with Udacity in the Udacity Data Science Nanodegree program.