Prediciton model

This topic has 4 replies, 3 voices, and was last updated 10 years, 9 months ago by Luciano.

Viewing 5 posts - 1 through 5 (of 5 total)

Author
Posts
November 11, 2009 at 11:34 am #26882

Muhammed
★★★★★★★★★★
Participant
@Muhammed Include @Muhammed in your post and this person will
be notified via email.
Hai iam Mohammed here,

Anybody could help me with a prediciton model since we are in the way of CMMI vesion 1.2 certification
0
November 12, 2009 at 10:15 pm #65350

Don Strayer
★★★★★★★★★★
Participant
@Don-Strayer Include @Don-Strayer in your post and this person will
be notified via email.
Go to the SEI website http://www.sei.cmu.edu and search for “self appraisal” to find some useful information. We used to do this at a former employer where I was an internal lead assessor, and found it to be a fairly accurate predictor. You must involve someone with thorough knowledge of the CMMI practices and how to interpret them. You must review documentation and conduct interviews. Decide whether or not each applicable practice is satisfied and write a gap analysis.
p.s. Avoid the term “certification”. SEI certifies appraisers and instructors but there is no certification of any sort for an organization’s CMMI level.
0
November 13, 2009 at 3:52 am #65351

Muhammed
★★★★★★★★★★
Participant
@Muhammed Include @Muhammed in your post and this person will
be notified via email.
Dear Don,
Thanks for your reply. i went through SEI site with the search key as “self appraisal” . But unfortunately i couldnt find any useful information over there. We are already in CMMI level3 and due to the business requirement we planned to moved Level5.So there i could found the prediction as the road blocker. Even thought we prepared some model for defect prediction, Effort variance. The problem what Im facing is to smoothening the appraisal need more models. So i was just looking other than theses two what models are using in the industry.
i really appreciate if you could help me with some sample prediction model.

0
November 14, 2009 at 4:57 am #65353

Don Strayer
★★★★★★★★★★
Participant
@Don-Strayer Include @Don-Strayer in your post and this person will
be notified via email.
Muhammed,
Try using dogpile http://www.dogpile.com It’s a meta search engine that compiles results from multiple search engines and helps you to use advanced search techniques to narrow the results.
There are plenty of tool vendors/consultants who’d be glad to sell you something but, frankly, you can do this yourself as long as you have or can contract someone who really understands L4 and L5.
0
November 18, 2009 at 8:46 am #65359

Luciano
★★★★★★★★★★
Participant
@Luciano Include @Luciano in your post and this person will
be notified via email.
Hi… I’m the metrics lead in a ML5 company.
My suggestion is:
1) First of all, connect things with strategy. Create a Balanced Scorecard (if you still don’t have one) and prioritize the processes that in your high level management’s opinion most impact the strategic objectives. Identify the main Ys that are critical to you (linked with strategy) and the xs (processes/subprocesses) that impact that Ys.
2) Describe THAT critical processes with one or two metrics each. Again, PRIORITIZE! Don`t try to model everything. 3 or 4 variables are hard enough to begin. After your first model is running, try to increase the variable number.
3) Collect them for a while (3-6 months if you don’t have historical data).
4) Use Minitab to ensure that the process is stable.
5) If it’s not or if you have bad data, don’t even try to create the model.
6) If data are OK and you have more than 100 data lines, try to create a regression model in Minitab.
7) R squared above 70% is nice to start…
0
Author
Posts