Let me explain what happens: given the form of the time series signal it is not always possible to find a model with a consistent bbehaviorto describe it.
In SAP BusinessObjects Predictive Analytics version 2.x (including version 2.4 that you are currently using) we used to give the message “The engine found no model” and stop the training.
In our next planned release - SAP BusinessObjects Predictive Analytics 3.0 we plan to use a default time series model (Exponential Smoothing). It might happen that such model gives equal values for all future forecasts - this is what happens for your data set. This reflects the difficulty of the time series modeling engine to find a “good enough” model.
I would strongly advise that you see if you can collect extra predictable variables. Please refer to the page 46 of our guide here: http://help.sap.com/businessobject/product_guides/pa25/en/pa25_ts_user_en.pdf for detailed explanations and an end to end scenario.
Can you please kindly close the post if your question is now answered?
Thanks & regards
Antoine