Capstone Project - The Battle of Neighborhoods


 

1         A description of the problem and a discussion of the background

 

We are planning to build a luxury hotel in Barcelona, and we want to know which neighbourhood is the best.

Barcelona was the 20th-most-visited city in the world by international visitors and the fifth most visited city in Europe after London, Paris, Istanbul and Rome, with 5.5 million international visitors in 2011. By 2015, both Prague and Milan had more international visitors. With its Rambles, Barcelona is ranked the most popular city to visit in Spain.

Barcelona as internationally renowned a tourist destination, with numerous recreational areas, one of the best beaches in the world, mild and warm climate, historical monuments, including eight UNESCO World Heritage Sites, and developed tourist infrastructure.

Barcelona is divided into 10 districts. These are administrated by a councillor designated by the main city council, and each of them have some powers relating to issues such as urbanism or infrastructure in their area. The current division of the city into different districts was approved in 1984. In 2009, in Barcelona started using a new division of 73 neighbourhoods (the 10 districts are still in use), a division that was done for a better service from the City Council.

Some of these districts have a previous history as independent municipalities which were integrated into the city of Barcelona during the late 19th century and the first half of the 20th century, such as Sarrià, Les Corts, Sant Andreu de Palomar, Gràcia or Sant Martí de Provençals. However, other municipalities which are contiguous to Barcelona (such as L'Hospitalet de Llobregat, Badalona, Sant Adrià de Besòs or Montcada i Reixac) have remained separate towns to this day and are part of the much larger metropolitan area of Barcelona.

We want to use an index of the family income (RFD) and the quantity of the hotels by neighbourhood to create clusters and see which one fits better with our goal.

2         A description of the data and how it will be used to solve the problem.

 

For this project we are going to use different datasets:

·         Barcelona’s last published income index by Neighbourhood (2017)

o    Source: https://opendata-ajuntament.barcelona.cat/data/dataset/79bdf758-dae1-485b-800c-be9f8cfa9360/resource/e7206797-e57b-4ded-8c6c-62e9b4cb54f7/download/2017_distribucio_territorial_renda_familiar.csv

o    Description: RFD (Renta Familiar Disponible per cápita) is the amount of income available to resident families for consumption and savings, once the depreciation or consumption of fixed capital in family economic operations and direct taxes and contributions paid to Social Security have been deducted. Eurostat recommends the use of the RFD as the main regional economic aggregate.

·         Barcelona Hotels Foursquare

o    Source: Foursquare API

o    Description: Using the Foursquare API we’ll get all the hotels in Barcelona.

·         Barcelona Hotels Open Data

o    Source: https://opendata-ajuntament.barcelona.cat/data/dataset/88efe464-2bcd-4794-85b0-8b0bbfd9e4c0/resource/eced0fe8-9892-4926-b035-4fdc7328e31d/download

o    Description: We’ve looked for another hotel’s source because we felt Foursquare API wasn’t complete enough. This data set includes an accurate description of every hotel in the city

 

3         Methodology

 

For this project we’ve use Open Data sources to know the family incomes by neighbourhood in Barcelona and for having a complete hotel’s list (to compare with Foursquare API) and the Foursquare API to find the hotels in Barcelona city.

We have done three different clusters to approach

3.1       Data Preparation

There is a Open Data Project in Barcelona so we use it to find the list of Barcelona Neighbourhood’s with its Name, Code, District, Population and incomes (RFD).

After some manipulation the first 5 rows are like this.

This data set doesn’t have the coordinates so we used Nominatim to Geocode the Neighbourhoods.

For the Hotels List we used two Data sources to compare. Here there is the preparation of the open data list.

After some manipulation we’ll have a new data frame with the hotels amount by neighbourhood.

3.2       Using Foursquare API

We have to use Foursquare because it’s a Project requirement so we prepared the Request with the required data.

Afther some manipulation, the result is:

3.3       Clustering

 

We have used three different approaches for clustering. All with kmean function but with different data

  •        With RFD Index, Population and Coordinates
  •        With RFD Index and Coordinates
  •        With RFD and the number of hotels


3.4       Data visualization

Barcelona with the 73 neighbourhoods

 

First Cluster

Second Cluster

Third cluster

4         Results

After some analysis we have found that the third cluster is the best who fits with our proposal, because it offers us a list of neighbourhoods with high income index and a low number of hotels.

 

 

There is the list of neighbourhoods that are in the Cluster number 3

 


5         Discussion

There is a list of observations for a future research:

  • Postal codes are not very useful in Spain because you can’t link it with the neighbourhoods.
  •   Foursquare API is not very useful in Spain because it hasn’t the venues with the required accuracy.

 

6         Conclusion

This was a Capstone Project so the risk of a bad interpretation is low compared with the required investment to build a luxury hotel in Barcelona.

We would avoid the use of the Foursquare API for a professional job if there is another data source available.

Comentarios