I'm trying out SSAS DAX Time Prediction-modeling in MSSQL 2012, but the forecasts do not match my expectation.
The input data is a perfect Sinus curve, about 300 datapoints with five repetitions 
. The data is loaded in the mining structure and mining model with time Microsoft_Time_Series and default algorithm parameters.
The prediction does not pick up on the Sinus-curve, but predicts an allmost linear line downwards.
What can I do to improve the prediction?
