Building a predictive model for government contractor rates using data from https://calc.gsa.gov/ which includes information from a vareity of GSA approved and managed contracting schedules. More than 80,000 labor rates are provided.
The focus of this project is to develop a useful predictive model to forecast government contractor rates for firms working with the US federal government.
This information can be useful to government agencies when developing Independent Government Cost Estimates (IGCE) for new procurements, and to government contrators when assessing the prevailing rates in the market for different types of services.
Steps completed to date (8/11/19)
- added features to data set for keyworks located in Labor Category Description
- built Random Forest, Decision Tree, and Linear Regression Models
- built scoring mechanism for all models (RMSE)
- built plots to show relative accuracy for all models
Planned next steps:
- parse SINs into additional features of data set, and train models with new information
- Add xgboost model
- Tune hyperparameters for RF and DT
- Look for additional information to refine data set...data on contractors listed in file?