I have one table in which price of some commodity over a time range of 4 years is available. In second table, I have a customer transaction history that tells me at what price customer purchased the commodity, no of days he kept the commodity, and at what price point he sold the commodity. Using this data I have to predict if that customer purchases the commodity in future at some price, then at what price and after how many days he will sell it.
I understand that the price of commodity can be predicted using time series analysis. However, I am not sure how to use this data to predict customer behaviour.
Now I need suggestion that on which factors does the customer behaviour depends i.e. does the customer sells a commodity if monthly rate of change is more than or less than a threshold or if there is some advance technique to predict customer behaviour pattern?