Francisco Marco-Serrano
20 November 2015
Competition can be addressed from a perspective of product differentiation. This differentiation can be considered in a multivariate facet.
In the economics literature there are models that consider the 'principle of minimum differentiation', determining competition by businesses differentiating themselves in one or few attributes. Here, we explore the following features:
Data on 945 Edinburgh restaurants from Yelp dataset has been used. See below statistic descriptives for the stars:
Min. 1st Qu. Median Mean 3rd Qu. Max.
1.500 3.500 4.000 3.806 4.000 5.000
The dataset can be obtained from: http://www.yelp.com/dataset_challenge
The analysis of the data has been done using random forests in an unsupervised mode.
The first 14 variables are related to schedule of the restaurant (open/closing hours); longitude and latitude -related to location- are in position 15th and 16th, with the review count being in position 17th. Quality, represented by the variable stars is 18th in the list of variable of importance to determine the clusters, followed by a list of several other attributes.
Restaurants in Edinburgh are competing in a first round, in terms of schedules, while location is of secondary consideration, followed closely by quality, in third position.
Using these results, further research could try to evaluate the changes in the competing scene forecasting the opening or closing of new restaurants in a defined location.