In this project, we conducted conjoint analysis to optimize product mix and redesign the product line for a small toy company, EarlyRiders.The datasets available were consumer rating data for different products and product profiles data
including information about mix of 4 attributes: price, height, motion, and style.
Tool: R
- did data cleaning and dealt with NA
- built a regression model to estimate each individual's part-utilities on 4 attributes and predicted for missing profiles (ratings and part-utilities)
- performed cluster analysis through K-means modeling based on conjoint part-utilities; visualized clustering results with pie chart, ellipse plot and bar plot; chose 3 as the number of segments based on elbow rule and plot reports
- conducted a priori segmentation using demographic variables, gender and age; constructed segment-level regression models to profile the attribute preference of each gender-age group
- simulated market shares and calculated profit for 14 different product line scenarios considering competitive response and cannibalization; recommended short term and long term product mix
Outcome: With our modified product mix, the profit will increase by 80% in five years.