481232_en-US

liliconcom

Re: Model Management Solution

A big wish indeed - Taking a look at SAS Model Manager shows that developing an out of the box solution likely to satisfy is 'not nothing'!

Re: Model Management Solution

One option we gave Regeneron is to check out CoBase from Predictum.  Predictum has also built out a model management and virtual lab for GoodYear, which is a former customer of mine.

Re: Model Management Solution

This is a really good idea to pursue. In the HLS space we have had some similar requests for features like these from Takeda, Sanofi, and some other smaller pharm companies. 

Re: Model Management Solution

I could see a solution like this being useful in fighting back against the python competition we're getting at Corning.   

Re: Model Management Solution

FWIW, I recently sat in on a demo of  CoBase from Predictum.  Predictum is currently redesigning CoBase and the updates, particularly the front end, presents very well to customers.

Model Management Solution

We recently had a discussion with a Data Science team that is part of the Preclinical Manufacturing Process Development group at Regeneron.  They are looking to put together a model management solution with the following functional criteria.  It would be nice to have an out-of-the-box solution that could put together such a solution (meeting most of these criteria) using JMP Pro and JMP Live

Functional criteria for model management solution

Area	Priority	Key Feature	Pain Point(-s)
Data Storage	High	Data changes history / Data versioning	No Version Control System
		Single data access point	No ability to trace data to the source
		Single storage for model documentation	No ability to trace data to the source
		Data Manipulation and Traceability	No ability to trace data to the source
			Data for modeling preparation is manual and data stored in multiple locations
Model Storage	High	Model Versioned Repository	No Version Control System
		Model Catalog	Model Search is difficult
Model Discovery	Low	Continuous model execution	Continuous/periodic model evaluation is manual and not regular
	High	Model Metadata Gathering	Difficult access to model meta-data across multiple models
			Model metadata gathering is manual
Model Manipulation	High	Model Comparison	Model comparison is manual
		Model Merge	Model merge is manual
Reporting	Low	Integration with reporting solution	Report creation is manual
Presentation	Medium	Across departments overview	No single-pane-of-glass view of models
		Historical models overview	No single-pane-of-glass view of models

 

 

Statistics and AnalyticsRe: Model Management Solution - Status changed to: Acknowledged

@andreacoombs1 thanks for passing this significant feature request along.  This has been captured in Jira and will be evaluated soon.  I appreciate the detail that you provided. 

 A couple of additional questions:

Do you why they want this?  What is the problem they are trying to solve or the tasks that they want to do?  

Do you know if this would result in significant value to them?  Can you quantifiy that?  

What are they using rignt now to do this?

Re: Model Management Solution - Status changed to: Investigating