Model Maintenance & Support

Data Science has changed significantly since starting in 2008. At first, data scientists were more concerned with the invention of the models, than implementing them. Companies (and the industry really) sought out inventors. Make no mistake, it takes invention to innovate; however, this had led to a bevy of models that have good r-squareds, but lack any kind of implementability inside the organization that built them.

Having R&D data scientists is a necessity, but, having only R&D data scientists (traditional data scientists) gives you a team that has no desire (actually unwillingly in some cases) to continually maintain their models. Once they finish a model, they want to move on to build and invent other new models. This is causing to 2 things:

1) models that are being created that solve real problems were not built in a way that can be maintained. They actually have to be rebuilt to be implemented into enterprise, and

2) rifts are developing that are causing exodus as companies need the data science team to stop building new and implement the models they have. (this is actually an incremental improvement problem). Data science teams have literally quit if asked to stop inventing and begin maintaining their models.

What to do?

In the same way that over half of model building is data gathering and manipulation of data, actually making a difference with the model over 50% implementation and maintenance. After all, the value of a model is zero regardless of the predicted ROI if it is never used.

CAN has developed a Model Maintenance and Support Program to solve this very problem. The MMS is a program designed to free your R&D data science team to do what they do best–invent, and give you the support and maintenance of the models needed to actually realize the predicted ROI.

The MMS program is a retainer contract, built around number and complexity of models–plus some dev hours for those unexpected issues.

This way, you get a level pay contract to maintain and support your models and piece of mind knowing a well-experienced data science team is taking care of everything.