Skip to content

Repo of several notebooks of research to the best methodology to use for predicting likelihood of student applicant-to-customer conversion. LightGBM ended up performing the best in terms of accuracy vs runtime.

Notifications You must be signed in to change notification settings

kenny-mai/student-yield-prediction

Repository files navigation

Subfolder for Yield Modelling

Steps to Reproduce

Virtual Enviroment

After cloning the repo, you will need to set up the virtual enviroment and install dependencies by running the following commands in the CLI in the folder:

  • python3 -m venv .venv
  • source .venv/bin/activate
  • pip install -r requirements.txt
  • pip install ../shared_packages/aws_helper_functions/

Setting Enviroment Vars & AWS Config

If run outside of lambda, applicable functions must be called with local_mode=True. Enviroment variables must be set for host, database, port, username, and password (eg redshift password) to connect to redshift. If writting results to S3 to update tables, AWS config must be set up w/access key and secret access key.

Files & Usage

  • yield_boosting.py -> entry point
    • train_basetable.sql -> query to create training data set
    • eval_basetable.sql -> query to create evaluation data set
  • requirements.txt -> packages that are requried to run intra_year_boosting (other than aws_helper_functions)

About

Repo of several notebooks of research to the best methodology to use for predicting likelihood of student applicant-to-customer conversion. LightGBM ended up performing the best in terms of accuracy vs runtime.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published