Opening a restaurant in Barcelona

The aim of my Data Science Capstone Project is to find the adequate neighbourhood in Barcelona to open a restaurant of a specific category, based on restaurants existing in a district. The idea behind the project is that people will tend to look for restaurants of a specific category in places where they know that there are restaurants of this kind. Barcelona has a growing tourist industry, and this analysis can be specially interesting for Barcelonian entrepreneurs looking to open a new restaurant venue. The template can be used for anyone that wants to explore the possibilities of opening a restaurant in a city.

I have used Foursquare location data to identify a sample of the restaurants existing in each neighbourhood. I have used Wikipedia to get the location of each neighbourhood, and then I have explored each location using the Foursquare app. Then I have filtered the locations with categories describing restaurants. I have obtained a sample of 1908 restaurants across several neighbourhoods of Barcelona. Then, I have obtained the proportion of each of the 58 categories of restaurants located in each neighbourhood, and I have used this information to cluster the neighbourhoods in different categories using k-means clustering. After testing several number of clusters, I have retained a clustering into three categories. The results of clustering can be viewed in the following map:

The most common restaurant category in  the red cluster is Tapas Restaurant. Tapas restaurants offer small plates (tapas, in Spanish) of Spanish food (e.g., patatas bravas). Tapas are always enjoyed in a context of leisure, either by tourists (in city centre neighbourhoods) or by locals (in the north neighbourhoods). Tourists can visit northern neighbourhoods (e.g. Horta or Turo de la Peira) for a more authentic tapas experience. This cluster includes 23 neighbourhoods, including the most touristic places in the center of Barcelona.

In purple cluster the most common category is Mediterranean Restaurant. These restaurants offer pizza and pasta, together with some Spanish specialities. These restaurants can be visited either by leisure or at a working pause at noon. These neighbourhoods have also a varied offer of international cuisine restaurants, sometimes owned by immigrant entrepreneurs.

The light blue cluster covers more than one half of Barcelona. In those neighbourhoods the predominant restaurant categories are Restaurant or Spanish Restaurant, followed by Mediterranean restaurant. People is going to these restaurants in the working pause. In Spain there is a tradition of making a long work stop (sometimes from 14 to 16) to have a strong lunch, with two plates and dessert. This is offered in Spanish Restaurants and Restaurants, and sometimes in Mediterranean Restaurants as menu del día. So we can expect that most of the revenue of these restaurants will be obtained in midday in working days, and that the leisure restaurant market is less buoyant in those neighbourhoods.

This analysis offers information about people's habits of interest for people wanting to start a restaurant business. This information must be complemented with information about real state market: for instance, Tapas bar can be opened more easily in neighbourhoods on the north of Barcelona belonging to cluster 1, where renting a place to open a restaurant can be cheaper. Futher analysis can be undertaken with the Foursquare app exploring trending venues in different moments of the week (weekends and working days) and of the day (working hours and night).

Human resource management and social network analysis

Human resource management research has been traditionally focused on individual and job attributes. Juman resource management research and practice could be furthered by studying attributes of relationships between individuals or jobs. Some fields where social network analysis is useful are (Hollenbeck and Jamieson, 2015):

• Identification and selection of employees. Social network analysis can help to predict future performance, for instance, by identifying what kind of employees plays brokerage roles across structural holes (i.e., links otherwise disconnected subgroups). It can also help to identify talent pools outside the organization, by identifying relationships of employees with external actors. Finally, it can help to identify isolated actors, which may have higher probabilities of turnover.
• Improving training and development programs. The outcome of training efforts depends of a strong network among employees. The identification of friendship and advice networks can help to improve communication, manage organizational culture and foster strategy implementation.
• Guide compensation and pay decisions by identifying key employees. Organizations must reward their employees on the results and behaviors that it expects from them. Most of this traits are socially driven (e.g., effective teamwork and prosocial behaviour, or integrating new employees). Social network analysis helps to recognize who is demonstrating these behaviours. 

References

Hollenbeck, J. R., & Jamieson, B. B. (2015). Human capital, social capital, and social network analysis: Implications for strategic human resource management. Academy of management perspectives, 29(3), 370-385.