New product Development w/ Conjoint Analysis
Ranking questions which represent preferences are converted into groups aka segmentes with similar preferences using KMeans cluster analysis.
Each observation (xi) is assigned to a given cluster such that the sum of squared (SS) distances of each observation to their assigned cluster centers (μk) is minimized.
The total within-cluster variation is defined as as follows:
In a further step, a nonlinear model (GLM) is estimated based on the choices collected through conjoint survey to obtain the partial utilities of product features for each segment defined by cluster analysis.
