One of the largest insurers in Brazil estimates gp effort based on a percentage of the global software development effort. It does this both for the purpose of planning performance in budgeting time, as well as in monitoring and the end of the project.
How to estimate GP effort in technical projects where % of the total does not work (e.g., procurement of packages or hardware)?
Solution
A spreadsheet, “calculator”, implementing a model of estimates of GP effort from the time, effort and other factors, which were discovered and whose impact on the GP effort was quantified.
Result
The average error, previously at 122%, fell to 36.6%; Only 25% of the initiatives had an error of up to 25%. With the model about 70% of the estimates presented an error of up to 25%
How the parts have been integrated
The Budget Center team analyzed documents and interviewed project management professionals in the client in order to understand their process.
Together with the client, held a workshop with the objective of identifying characteristics of a project of this nature, which promote a greater or lesser impact on the effort of INVESTED GP.
This material allowed the elaboration of a preliminary data collection structure with 17 factors, validated with selected professionals in the client. 25 projects were selected and 10 GP received a kit with a collection form to evaluate their projects.
The Analytics and Machine Learning team used data analysis and statistical modeling techniques, using the RStudio platform as support, enabled them to simplify in only 5 input parameters and design an estimate model implemented in an Excel spreadsheet.
