Four different ways to see an urban space through data. Image based on a photo by Kaique Rocha from Pexels
The Data-driven approach encourages the use of the information generated in urban spaces to get valuable knowledge through the right analysis of this information. This analysis not only allows us to know what is happening at any given time but also lets us to learn from the past to take preventive measures and act efficiently in the future.
Within the data analytics field there are four different types of analysis, each one of them answering to a generic business question:
- Descriptive analysis: What is happening?
- Diagnostic analysis: Why is it happening?
- Predictive analysis: What is likely to happen?
- Prescriptive analysis: What should I do?
Now then, how can take advantage an urban space that follows the Data-driven approach of these kinds of analysis? The answer, below:
Descriptive analysis
This kind of analytics presents objective, accurate and real-time information from primary sources such as sensors, cameras or social networks that can be minimally processed. The key point of the descriptive analysis is to offer an effective and comprehensive visualization of what is happening in an urban space, as if an image were taken in a certain moment and, of course, this image can be visualized later in case it is required.
The image above shows a screenshot of the map of the real-time traffic conditions in the city of Barcelona the Thursday 31st January, 2019 at 12pm, get from the Barcelona City Council website, in which we can see there are some congested road sections.
Diagnostic analysis
This kind of analytics goes deeper in the situation observed while isolating confusing information. In this case the key point is the elaboration of hypothesis as well as crossing data from several sources to corroborate these, always keeping in mind that correlation among attributes does not imply causation.
Which could be the cause of the congestion seen in the map shown in the image above? It could be a punctual incidence such as an accident or a damaged vehicle or not properly parked, or some works in this section of the road or even a demonstration that took place there at that time. Maybe we are facing a systemic situation that is produced every day at the same time at this point, or that is produced when some meteorological conditions are met. Whatever the hypothesis is, it should be checked with the right analysis of the data.
Predictive analysis
This type of analytics uses of the previously gathered data to establish patterns according to Machine Learning models so that we could predict what is likely to happen in the future. Decisions based on these analysis can be automatized with a high degree of reliability, however it must be taken into account that the values of the predictive variables used are within the ranges in which the model was trained, otherwise the accuracy of the obtained results cannot be fully guaranteed.
Once identified the attributes that most influence traffic conditions in the city, it is possible to create a model that predicts circulation fluency in a given road section. Let us suppose a very simple model that only took into account the weekday, the time and weather conditions to determine traffic status; given a Thursday at 12pm in a rainy day this model will tell us that road conditions on Aribau street would be dense.
Prescriptive analysis
The next step in terms of value and complexity is the prescriptive analysis. This kind of analytics applies advanced analysis techniques that consider information about what happened before, the causes, as well as some “what if” hypotheses to make some recommendations that facilitate the election of the most convenient actions all times.
A good example of prescriptive analysis, following the line of the previous examples, would be a driving assistance application which would help users to choose the best route to their destination, among several alternatives, taking into account traffic conditions in real time.
Conclusion
Data analytics offers many possibilities to manage the urban spaces of the 21st century. The Data-driven city has powerful tools to anticipate all those situations that may occur there and facilitates the management of daily activities as well as those specific facts potentially problematic that could negatively affect normal life in cities. Each one of the four kind of analytics introduced in this post brings great value to urban managers to ensure the optimal performance of their city and its services.
The Bit City 