A well-known data mining process standard called Cross-industry standard process for data mining (CRISP-DM) [1] is an open standard process model that is utilized to describe the major building blocks of data mining processes. It is the most widely-used data analytics model [2] and describes common approaches employed by data mining experts in their data analytics projects. As shown in Figure below, CRISP-DM consists of six phases: 1) Business Understanding (the phase in which you get an understanding of the problem going to solve, its impact on your organization and generate project plan); 2) Data Understanding (the phase in which you inspect, describe and evaluate the data that will be used to solve the problem); 3) Data Preparation (the phase in which you prepare the data in a format it is needed for analysis); 4) Modeling (the phase in which you apply mathematical models and algorithms that facilitate informed business decisions); 5) Evaluation (the phase in which you test whether your model is performing well); and 6) Deployment (the phase in which you integrate the newly developed model into the business application).

My question is, why shouldn’t we connect Deployment phase back to Business Understanding? In my view, once a model is deployed, various unforeseen issues may arise in the real-world application in which our model is deployed. That could impact the business in various ways – some of which could be contrary to our finding in the initial business understanding phase.

Refs:

[1] Shearer C., The CRISP-DM model: the new blueprint for data mining, J Data Warehousing (2000); 5:13—22

[2] What IT Needs To Know About The Data Mining Process Published by Forbes, 29 July 2015, retrieved June 24, 2018