Predictive modeling contains six stages of analysis according to the Cross Industry Standard Process Model for Data Mining (CRISP-DM). I will break this down into three primary tasks for predictive modelers including preparing data, building models, and explaining results.
Data preparation often requires skills in SQL, python, or other languages to be able to pull data out of data stores and convert the normalized data into flattened data that the algorithms can use to build models
Modeling requires a qualitative (if not quantitative) understanding of the algorithms, including mathematics or statistics, in order to build effectively.
Finally, modelers should know how to explain the results of their findings to other analysts and to decision-makers and stakeholders.
The session will walk through the building of a predictive model for a retail application: predicting the days to next purchase propensity model.
No material found.