Summary:

• Data fragmented into 4 Segments using a combination of Principal Component Analysis (4 Components) and K-Means Clustering.
  1. Career-focused - Young group of high-income and well-qualified people.
  2. Well-off - Mid-Age wealthy group in a big city.
  3. Standard - Average Population. (Middle Class)
  4. Few opportunities - Customers from small cities with mediocre income and not highly qualified.
• A Descriptive Analysis of the Number of purchases, Brand choice, and Revenue generated per segment.
• Price Elasticities (Modelled the elasticities for each price point starting 0.5 to 3.5, for 300 values.)
  • Purchase Probability -
    • Customers from Well-off & Career-Focused segments are less elastic.
    • Customers are less sensitive to price changes when there are promotions.
    • Beneficial to have a higher original price and constant promotions.
  • Brand Choice Probability (cross-brand) -
    • Brand 5 can reduce its price by 1% to gain its market share from Well-off segments.
    • Career-Focused Segment are loyal customers of Brand 5.
    • Brand 1 has to reduce its price by 2% to gain some market share from Brand 2 in the Standard and Few Opportunities Segments.

Resources

• Python Version: 3.7
• Packages: pandas, numpy, sklearn, scipy
• Visualizations: matplotlib, seaborn
• Techiques: Linear Regression, Logistic Regression, K-Means Clustering, Principal Component Analysis
• Full Code can be accessed at the Github repository