This project is focused on training a Linear Regression model to predict future Return On Investment (ROI) for various advertising spend budgets across search, video, social media, and email channels.
Tech stack includes -
- Streamlit
- Snowpark
- Plotly
- Tensorflow
- ML LR model training code on Snowflake using Python Stored Procedure
- Scalar and Vectorized Python User-Defined Functions (UDFs) for inference
- Snowflake Task to automate (re)training of the model
- Streamlit web application that uses the Scalar UDF for real-time inference on new data points based on user input
- Cohort Analysis on sales data for month on month sales w.r.t. unique users
