Predicting UnitSales given the price

emoscosocamemoscosocam Member Posts: 1 Contributor I

I am new with this tool and with Data Mining in general. I have a possible application:

I have the following historical data from a retail store:

(#Year, #Month, #DayType,  #ProductCode, ProductCategory, AveragePrice, UnitSales)

DayType is a number from 1 to 3, where Monday and Tuesday equal 1, Wednesday and Thurday equal 2, and finally Friday, Saturday and Sunday equal 3. AveragePrice is decimal number, UnitSales is integer. So, the historical data shows us how many units has sold a given Product that belongs to a certain ProductCategory in a DateType as a result in a change of the AveragePrice.

If I have a new product that belongs to a certain ProductCategory, I would like to se Rapid Miner in a way that it predicts the UnitSales given the Price and other parameters, something like:

UnitSales = PredictionFunction(ProductCategory, Month, DayType, Price)

What suggestions may you give me to solve this?

Thanks a lot.


  • venkateshvenkatesh Member Posts: 15 Contributor II
    What you want is to predict the performance of a product category in the coming week. It depends a bit on how many previous weeks data you want to use to predict next week's. What you have is time series data, you can use time series extension of RapidMiner, take a look in to windowing. Windowing is stiching n weeks of data as a row, based on that you can predict next week performace, It becomes a simple classification problem
Sign In or Register to comment.