Modern Business Analytics

Chapter 2. Methodologies for the Business Analyst and Analytics Projects

Analytic projects are unique in that these projects produce working software. Within the industry, this software is referred to as a data product. Data products are those outcomes of analytics that are used by an organization as part of business operations. A common example of a data product would be a predictive model. For instance, this could be a model that predicts the likelihood of a customer clicking on a pop-up ad. The methodology that is the basis for most analytical projects today is called the Cross Industry Standard for Data Mining (CRISP-DM). While it is not referred to any longer due to its age and more modern methodologies being introduced, the modern methodologies still use the primary phases from CRISP-DM. Each phase has a purpose and outcomes that business analysts often participate in. Figure 2-1 highlights the phases and possible iterations an analytics project might take.

Throughout this chapter, we’ll dive into the six phases of an analytics project, which are business understanding, data exploration and preparation, modeling and evaluation, and deployment. We will also explore what happens to a data product once used as part of business operations. In the technology world, a data product used in business operations means that the software that produces the data is “in production.” When software is in production, it requires support and monitoring, and this is referred to as model operations. ...