Title-Yelp AZ Restaurant Reviews

Introduction: The following is an analysis of business reviews provided by Yelp.com.

Audience: Restaurant owners in Arizona. Question: What should I avoid doing and what extra should I consider doing in order to receive good customer reviews?

Goals: Account for user bias, find terms we can measure on and score each of those terms. The next slide describes how the goals were addressed.

User bias is addressed by using the review Z-score of each user. We use Food Menu Items as the terms to quantify the relevance of each.

Using a binomial Generalized Linear Model for each Menu Item. Here's an excerpt of the code:

search.glm <- glm(current.matched[,j]~zscore, family=binomial())
search.predict<-predict(search.glm, data.frame(zscore=c(-2, -1, 0, 1, 2)), interval="link", se.fit = TRUE)

The plot summarizes the change in odds as the Z-score increases. More analysis available in the report.

• Avoid doing - Serving any item on the first Results plot that scores under 1. If that can't be avoided, like for Water or Chips, I would recommend a laser focus quality control on these.

• Don't bother - Caesar Salad, Spinach Dip, Guacamole and Pulled Pork they do not seem to improve the odds of getting a good review.

• Consider doing - In order to improve chances of getting a good review, restaurants in Arizona should consider serving: Cannoli, Creme Brule, Tiramisu, Ravioli, Gnochi, Pancakes, Brisket, Hot Dogs and Soup. I would absolutely feature and advertise Pizza.