This content is part of the Essential Guide: How to build an AI services business in the channel
Manage Learn to apply best practices and optimize your operations.

Cognilytica expands on CRISP DM model for AI project management

What do you get when you cross the CRISP DM model with Microsoft’s TDSP?  Well, yes, another acronym. But Cognilytica, a market research firm in Washington, D.C., believes the pairing provides a methodology for AI project management.

CRISP DM stands for Cross-Industry Standard Process for Data Mining, a step-by-step approach for launching a data mining project. CRISP DM was created in the late 1990s, before the current surge in AI investment. TDSP, or Team Data Science Process, is a methodology for implementing data science initiatives. Microsoft introduced TDSP in 2016, with the aim of facilitating predictive analytics offerings and intelligent applications.

Cognilytica’s Cognitive Project Management for Artificial Intelligence (CPMAI) methodology combines CRISP DM, TDSP, Agile and its own thinking and research on best practices for AI project management. The company offers training and certification on CPMAI.

Building upon the CRISP DM model

Ron Schmelzer, managing partner and principal analyst at Cognilytica, said AI projects call for a different take on project management compared with traditional IT initiatives. Specifically, an approach for running an AI, machine learning (ML) or cognitive technology project must be “much more data centric,” he noted.

The CRISP DM gets an AI adopter partway there. The methodology starts with understanding a business’ data mining goals and works its way through phases such as data collection, preparation and modeling. CRISP DM, however, “is not AI-specific and is missing some AI details,” noted Kathleen Walch, managing partner and principal analyst at Cognilytica.

CPMAI takes the data centricity of the CRISP DM model and adds TDSP, which Microsoft describes as an “agile, iterative data science methodology.” The Team Data Science Process encompasses a data science lifecycle definition and a standardized project structure along with infrastructure, resources, tools and utilities.

Cognilytica, meanwhile, contributes components such as best practices from in-production AI implementations, ML model training approaches and ML model evaluation.

Broader AI discussion

Schmelzer and Walch discussed CPMAI during last months’ AI World Government conference in Washington, D.C.  Other conference speakers outlined a three-step process for launching AI projects and discussed conversational AI’s potential in the public sector.

AI World will take a wider-angle view of the technology when the conference convenes Oct. 23-25 in Boston. The event will consider AI in manufacturing, healthcare, pharmaceuticals and financial services among other markets.

Cloud Computing
Data Management
Business Analytics