Have you experienced or thought how corporates manage their analytical assets which are mission critical to the business? A Bank or a Telecom Service Provider may often have more than 100 predictive model assets developed over a time period, but faces an important issue of how to effectively manage,store,share or archive these assets.

The next breakthrough in data analysis may not be in individual algorithms, but in the ability to rapidly combine, deploy, and maintain existing algorithms.”

Many corporates have now realized the need for a centralized repository of storing predictive models along with detailed metadata for efficient work-group collaboration and version control of various models.This blog is about understanding the underlining idea about model management tools.

Model management involves a collaborative team of modelers, architects, scoring officers, model auditors and validation testers. Many organizations are struggling with the process of signing off on the development, validation,deployment, and retirement life cycle management milestones. They need to readily know exactly where each model is in the life cycle, how old the model is, who developed the model, and who is using the model for what application.The ability to version-control the model over time is another critical business need which includes event logging and tracking changes to understand how the model form and usage is evolving over time.

Model decay is another serious challenge faced by organizations. Metrics are needed to determine when a model needs to be refreshed or replaced. Retired models also need to be archived. More reliable management of the score repositories is also a key requirement to ensure that quality representative data is available to evaluate model performance and profitability over time.

Two such tools that provide model management capabilities are SAS Model Manager and SPSS Collaboration & Deployment.

SAS Model Manager is designed for selection,maintenance and continuous enhancement of analytical models for operational decision making.This tool enable process to effectively manage and deploy analytical models by delivering all the necessary functionality for each stage of model life cycle. A typical Model life cycle has 4 stages namely as depicted in the figure below.

SPSS Collaboration & Deployement tool also provide similar functionality as of SAS model Manager tool.

Below URL shares the features of both these tool.

Views: 6331


You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by Adrian Walker on October 3, 2016 at 1:22pm

Models written in Executable English can be much easier to manage.  They explain their results.  Google indexes and retrieves them. 

Here's a slide:


and a short paper:


The system is live online at executable-english.com  -- shared use is free, and there are no advertisements.

Comment by nauman sheikh on April 2, 2015 at 4:28pm

My book "Implementing Analytics: A blueprint for Design, Development and Adoption" that I wrote in 2012 and published in 2013 has entire chapter with a design pattern how to build a model management sub system. You don't need to buy expensive software package to be able to track the performance of models, review their effectiveness and tune or replace them with proper checks and balances.

A simple sub-system with 10 or so tables can be built with a simple set of screens to achieve the same goal. The book has a detailed layout how and what such a model management sub-system should do.

© 2021   TechTarget, Inc.   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service