Description du stage:
Business knowledge and expectations can help us formulate opinions on how the regression coefficients should look like when applying this approach to generate forecasts. For example, some coefficients need to be negative while other coefficients should not be greater than a certain threshold. The objective of this project is to investigate an implement an approach to include an arbitrary set of constraints that will be taken into consideration when generating the regression cofficients. The algorithms need to be efficient to yield results within acceptable time limits given relatively big input data sets.
The intern will be required to:
- Transform the raw data to be used by the regression algorithm
- Run the constrained regression
- Assess the regression results and compare them to previous results of other approaches
Spark, Scala, Python, any other technologies that the intern deems necessary