Process Overview
From Maisqual Private Wiki
This article explains how information is gathered, analysed and transformed.
[edit] Process Overview
We consider information as a suite of indicators evolving across versions/releases.
One of the first thoughts was to set process information and product quality as inputs and outputs of the statistical model. This was not right because many base measures are relevant for both practices identification and quality measurement, which is statistically problematic: inputs and outputs shall be independent for the algorithms to work properly.
Rather than trying to correlate end-purpose artefacts, which would lead us to correlate the way we build models and introduce a bias, we want to correlate the base measures only. Given these results, we can then build practice and quality models from different aggregation methods.
At the moment, the sampling frequency is set to the release at the moment, but it could be lowered to some fixed-length periods, e.g. weeks or days, to allow fine-tuned advice during the iteration.
[edit] Base Measures
After data retrieval, we have the following set of indicators:
Project 1.0:
| Project 1.1:
| Project 1.2:
|
The base measures we want to retrieve have been gathered in the Base Measures category.
[edit] Quality and Practices models
All base measures are then combined into:
- quality attributes, which allow to measure quality of product and process,
- practices, which allow to know how the project is managed.
Model A:
Product Quality:
| Process Performance:
| Charisma:
|
Model B:
Product Quality:
| Process Performance:
| Charisma:
|
From the data retrieval and analysis point of view, we clearly differentiate base measures and models, so several different quality and practice models can be applied to the same set of indicators. We really want to correlate base indicators, not the way we build the models.
Project 1.0:
| Project 1.1:
| Project 1.2:
|
