This is sort of related to a previous question, but now I don't have the requirement of generating customer-level forecasts.
I've acquired a set of card customers every month for the last 3 years. I want to generate 24-month forecasts for the my most recent 3 cohorts (which have 1,2,and 3 months of spend data).
My data is at the customer-level, but I can aggregate it up to the cohort level.
I've essentially boiled down the problem statement to "generate a time-series forecast with little historical data, but where I have other past similar time-series".
I've heard of two potentially promising techniques and was hoping others could point me to either other techniques I can explore or help me understand whether the following techniques actually are promising:
- Panel Data Regression
- Bayesian Time-series structural models
Thank you; let me know if I can provide more insight.