A consumer is faced with x voucher categories for services located in y geographical regions. Having a history of consumption for this user over time T, what statistical model should be used to derive predictive consumption preferences?
If your Groupon-like voucher categories do not have some kind of natural ordering, i.e., they cannot be ranked, I would go for multinomial logistic regression.
Are interested in change over time X region? Do you want to know if the consumer changes focus to different areas (are they related to distance or is your consumer moving into different areas)? Suggest to map to see if there is a pattern.
Is your consumer data independent over time (eg. do the consumers buy a new product category, than change to a different category because they are bored with the old one or do they buy the same product category from different areas)?
You might need to check for autocorrelation and such.
Check out http://www.statoek.wiso.uni-goettingen.de/veranstaltungen/zeitreihen/sommer03/ts_r_intro.pdf
http://a-little-book-of-r-for-time-series.readthedocs.org/en/latest/src/timeseries.html#time-series-analysis
I would suggest multinomial regression as behavior is involved which can't be ranked
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series