Models

Model 1: Baseline Linear Regression The formula we use to predict a restaurant’s rating in our baseline model looks like this:

Ratingit​=β0​+β1​⋅Pricei​+β2​⋅Deliveryi​+β3​⋅Takeouti​+β4​⋅log(Reviews)i​+εit​

Rating: This is the predicted star rating for a specific restaurant.

β0: This is the baseline rating for a restaurant, before considering any other factors.

β1: This represents the adjustment to the rating based on a restaurant’s price level.

β2: This represents the adjustment to the rating based on delivery availability.

β3: This represents the adjustment to the rating based on takeout availability.

β4: This represents the adjustment to the rating based on the restaurant’s review volume.

εi: This is a small adjustment for all the random factors that the model doesn’t account for.

Model 1 is a straightforward linear regression that predicts restaurant ratings based on four key characteristics. It assumes that the relationship between these factors (price, delivery availability, takeout availability, and review volume) and ratings stays constant over time. This model gives us a baseline understanding of what generally makes restaurants successful, but it doesn’t account for how these relationships might have changed during different periods like the COVID pandemic.