Great article, I think it will really help people new to data mining models in SSAS.
An important point: you said in the beginning "we need a lot of data to predict the future"
and then in the end of the article reaffirmed that you had predicted the future.
No amount of data will predict the future, and those people developing data mining models have a responsibility to ensure that their clients don't believe that they can magically predict the future. All you can do is make predictions within probability, and recognise that the probability value is only as good as the training data. The more data you feed the model, the closer it will converge on its true probability value which is not the same as converging towards 100% accuracy.