Data Ingestion & AI Feature Engineering

Reading the OTT viewership dataset and checking the raw business, marketing, timing, and audience activity fields.

Preparing the timing-related inputs and using Generative AI to create the risk_score feature.

Building an enriched dataset that contains both the original variables and the new AI-generated feature for regression modeling.

Baseline Regression Model & LLM Interpretation

Training the first Linear Regression model using all available features in the dataset.

Evaluating the model and capturing the coefficient table, p-values, and performance metrics.

Using Generative AI to interpret the regression coefficients and explain the key business drivers in simple, business-friendly language.

Refined Regression Model & Model Comparison

Identifying the statistically significant predictors from the baseline model and removing the weak or redundant variables.

Retraining a simplified Linear Regression model using only the most important features.

Comparing the refined model with the baseline model to check whether we can achieve nearly the same performance with a simpler and more interpretable workflow.

ShowTime AI-Driven OTT Content Viewership Prediction