Data Science Methodology

Data Science has a great potential to transform organizations and to produce outstanding results. What are the steps needed in Data Science projects to increase the likelihood of success? One approach which has stood the test of time is the CRISP-DM methodology (CRoss Industry Standard Process for Data Mining).

CRISP-DM was established in the late 1990s and partially funded by the European Commission under the ESPRIT Program. Contributors included:

• NCR Systems Engineering - later spin off Teradata
• SPSS, Inc. - acquired by IBM
• Daimler Chrysler
• OHRA Verzekering en Bankk Groep

The methodology uses an iterative approach, where some steps can be revisited, as needed. Data Science is often a learning process which requires changes as efforts take place. This diagram depicts the CRISP-DM workflow.

A separate, detailed tutorial about the CRISP-DM methodology is under development. Check back again soon.