Data visualisation is a vital technique in which to display data insights in a way that their implications and importance will be understood. Here at Forecast, we are always interested in the best way for our clients to receive our work and insights. Power BI is a cloud-based analytics service, with the aim to provide interactive visualisations and insights from a variety of data sources. A report which collates all the information is first created, then can be published as an online SaaS (Software as a Service), to be viewed online on desktops or mobile applications. An example of what this looks like can be seen in the sample report below, which analyses car accidents in Edinburgh between 2009-2014 (data downloaded from Kaggle) to make data driven recommendations to improve road safety.
Our report shows some exploratory data analysis on the dataset on the first page, which in this case contains information such as the time, co-ordinates and severity of the accidents. From this, we chose to concentrate on the month of August, as this was the month with the most car accidents in our time period. This is most likely the case because of the Edinburgh Fringe Festival (the world’s largest arts and culture festival), hence a sharp rise in the number of visitors to the city.
We then looked at the day of the week, time and location of the accidents on the following page, to observe any patterns between the accidents and the number of casualties with location. The interactivity of the report can be explored here, where the client can click on a particular year in the selection box on the left. If required, specific points on the other visualisations can also be examined by hovering over visualisations. A click of a data point then filters the other visualisations from the chosen selection. The data in this report shows that most of the car accidents in August are happening in the afternoon, usually in the second half of the week.
The visualisations continue onto the next page, which highlights the type of pedestrian crossings involved in the reported accident, as well as the number of vehicles. Again, the selection box on the left can be used, or single data points can be selected for filtering. This page shows that most accidents include only one vehicle, and mostly where there is no crossing within 50 metres or during the pedestrian phase at the traffic lights. This indicates that these accidents likely involved a pedestrian with the vehicle.
From this type of analysis, recommendations on how to improve road safety in Edinburgh’s City Centre from our demo scenario can be made. In this case, it could be to include more pedestrian crossings or other types of traffic control on the more accident-prone streets during the Fringe festival in August. Traffic could also be diverted away from the City Centre, particularly at this time of year, and more information for tourists could be made available on road safety and pedestrian crossings.
