This section will look at each of the analytics methodology steps individually. I will use CRISP-DM as the template, because it covers model deployment, and we have already mentioned the benefits of sampling (which is the first step in SEMMA).
Many predictive modelers assume that the actual modeling phase is where the most insightful model development takes place. However, much of the groundwork and insight can be discovered early on, and a good understanding of business objectives can avoid pitfalls later on.
I must admit, business people and technical people can be better at communicating with each other. How business goals are communicated can run the gamut. It can be anything from a business partner stating, "Tell me how sales need to be increased" or "Tell me something I don't know."
So, it really starts with understanding what the specific business objectives are...