Choice-Based Conjoint Analysis
A notebook-based study of choice-based conjoint analysis, from choice-task design to utility estimation and market simulation.
Problem
How can stated choices be translated into estimates of preference for subscription attributes and their levels?
This project applies Choice-Based Conjoint Analysis (CBC) to Netflix-style subscription choices. The notebook is the primary artifact. It documents the analysis workflow; it should not be read as a representative market forecast without information about sampling, questionnaire quality, and model assumptions.
What is Conjoint Analysis?
Conjoint analysis answers the critical business question:
“Which features of a product are most important to customers, and what trade-offs are they willing to make?”
Instead of asking customers to rate features individually, conjoint analysis presents realistic choice scenarios—mirroring real-world purchasing decisions.
Key Concepts
| Concept | Description |
|---|---|
| Attributes | Product features being evaluated (e.g., price, content library, video quality) |
| Levels | Possible values for each attribute (e.g., price: $9.99, $14.99, $19.99) |
| Part-Worth Utilities | Numerical values representing the desirability of each level |
| Choice Sets | Combinations of attribute levels presented to respondents |
Methodology
1. Survey Design
Design choice tasks where respondents select their preferred option from realistic product configurations.
2. Data Collection
Use survey responses that contain a selected option for each choice task. The validity of any market interpretation depends on the actual sample and survey design.
3. Statistical Modeling
Use multinomial logistic regression to estimate part-worth utilities for each attribute level.
4. Market Simulation
Predict market share for different product configurations based on the estimated utilities.
Interactive Analysis
Explore the complete conjoint analysis workflow, from data preprocessing to utility estimation and market simulation:
Loading notebook...
Business Applications
-
Pricing analysis: Compare the preference trade-offs associated with different price levels.
-
Feature prioritization: Examine which attributes contribute to modeled utility.
-
Scenario analysis: Simulate configurations under the assumptions of the fitted model.
-
Segmentation: Explore whether preference patterns differ across defined groups.
Product value and limits
The notebook shows how choice tasks become utilities and simulated configurations. That gives a product team a traceable way to discuss trade-offs instead of relying only on isolated feature ratings. It does not by itself establish population-wide preference, an optimal commercial bundle, or causal behavior. Those conclusions require validated survey data and a documented sampling design.