- PraveenFor the moment, let us ignore issues wrt PPMsIf you have only one data point, how would you extrapolate future performance?Ed----- Original Message -----
**From:**praveen**Sent:**Saturday, July 31, 2010 7:44 PM**Subject:**[CMMi Process Improvement] PPM related questions / helpHi Group Members

I seek your expert advice on some of the questions related to PPM. I am helping an organization to implement PPM. Their team has defined PPM for:

1. Defect Removal Effectiveness

2. Customer Satisfaction:

3. Schedule Variance

These PPMs are dependent on the following Xs:

1. Total Efforts (Continuous) : 2 or 3 data points will be available if re-estimation is done

2. Team’s Domain Knowledge (Discrete: Y / N or 1 / 0)

3. Re-planning Done (Discrete: Y / N or 1 / 0)

4. Overall Review Efficiency (for this only one data point will be available)

5. Proactive action taken to solve any customer issue (Discrete: Y / N or 1 / 0)

Like this there are some more variables (X) are considered which are:

1. Discrete

2. Available one time

3. Not under statistical control

My questions are:

1. Is it OK if Xs are not under statistical control ?

2. Is it fine to use discrete variables for predictions ?

For example: In case of Defect Removal Effectiveness, it depends upon overall review efficiency (available one time), Domain knowledge of team (discrete), efforts etc.

In this case, there is very little scope of changing / control any of the Xs. Also number of data points available will be very less. In some cases only one data point will be available.

Team’s argument: they have followed same approach as mentioned in one of the training material of SEI, which mentions about usage of discrete variables for prediction.

Please help me in getting some more insight into this.

Thanks in advance. Praveen

- PraveenWith only one data point, future performance can be extrapolated in any direction - all 360 degrees, unless the physical world prohibits going backwards in space or timeSuggest you start with http://en.wikipedia.org/wiki/Extrapolation -if the team does not understand this you are in a bit of trouble as they do not understand some basis elements of mathematics and/or the real world.Ed----- Original Message -----
**From:**praveen**Sent:**Saturday, August 07, 2010 9:54 AM**Subject:**Re: [CMMi Process Improvement] PPM related questions / helpHi EdI agree with you. But how to convince the team. Can I show them some examples?Any one can share some more info??

Thanks in advance.

Praveen

--- On**Sun, 8/1/10, EDWARD F WELLER III**wrote:*<edwardfwelleriii@ msn.com>*

From: EDWARD F WELLER III <edwardfwelleriii@ msn.com>

Subject: Re: [CMMi Process Improvement] PPM related questions / help

To: cmmi_process_ improvement@ yahoogroups. com

Date: Sunday, August 1, 2010, 7:03 PMPraveenFor the moment, let us ignore issues wrt PPMsIf you have only one data point, how would you extrapolate future performance?Ed----- Original Message -----**From:**praveen**Sent:**Saturday, July 31, 2010 7:44 PM**Subject:**[CMMi Process Improvement] PPM related questions / helpHi Group Members

I seek your expert advice on some of the questions related to PPM. I am helping an organization to implement PPM. Their team has defined PPM for:

1. Defect Removal Effectiveness

2. Customer Satisfaction:

3. Schedule Variance

These PPMs are dependent on the following Xs:

1. Total Efforts (Continuous) : 2 or 3 data points will be available if re-estimation is done

2. Team’s Domain Knowledge (Discrete: Y / N or 1 / 0)

3. Re-planning Done (Discrete: Y / N or 1 / 0)

4. Overall Review Efficiency (for this only one data point will be available)

5. Proactive action taken to solve any customer issue (Discrete: Y / N or 1 / 0)

Like this there are some more variables (X) are considered which are:

1. Discrete

2. Available one time

3. Not under statistical control

My questions are:

1. Is it OK if Xs are not under statistical control ?

2. Is it fine to use discrete variables for predictions ?

For example: In case of Defect Removal Effectiveness, it depends upon overall review efficiency (available one time), Domain knowledge of team (discrete), efforts etc.

In this case, there is very little scope of changing / control any of the Xs. Also number of data points available will be very less. In some cases only one data point will be available.

Team’s argument: they have followed same approach as mentioned in one of the training material of SEI, which mentions about usage of discrete variables for prediction.

Please help me in getting some more insight into this.

Thanks in advance. Praveen