Way too many factors - Factorial Reduction
EXAMPLE:
Owner of the 'DVD-movies' shop asked customers to participate in the survey
to rate their liking of different movie types.
Survey range in value from 1 to 5, which represent a scale from Strongly Dislike to Strongly Like.
TESTING - Dimention reduction:
All factors are selected and proceeding with Principal components method for dimension reduction.
Correlation between movies types were found. All are above .3 which is good.
Two components that were created explains:
81.9% of the variance in 'Comedy',
76.8% of the variance in 'Romance',
82.2% of the variance in 'Family',
79.1% of the variance in 'Action',
84.5% of the variance in 'Horror',
and 52.1% of the variance in 'Thriller'.
There will be some information loss as factors are reduced from 6 to only 2 components.
RESULTS:
Only components with Eigenvalue > 1 are taking into the model.
For 2 created components the cumulative variance is 76%.
Which means that those 2 components explain 76% of original data.
EXPLANATION:
Shop owner understood that his clients are divided into two larger groups.
First one, is more likely to choose family or romantic comedy movie, while second
will rent scary thriller full with action elements.
