Peter Fader, a quantitative marketing professor from Wharton, believes that companies are unnecessarily complicating the job of making predictions. Simple probability models, often built in excel, can be far more powerful in many cases than complex statistical algorithms. He also believes that businesses can get by with much simpler and more powerful analytics if they stop worrying about why something will happen (which is a lot harder to answer) and just focus on what is predicted to happen.
“What Data Mining Can and Can’t Do”