Data-Driven Optimization For Shared Mobility Solution

One of the key factors in managing shared mobility platforms is knowledge about the demand for vehicles in place and time. This knowledge gives the company the ability to plan the distribution of vehicles in the city to better serve customer demand, as well as realise marketing strategy more effectively.

Equally important is to get to know your customers – their needs, habits and preferences. Thanks to that, shared mobility companies can enhance their service and user experience, create personalized special offers to improve sales and decrease customers’ retention.

The aim of this project was to provide our client with knowledge about the vehicles demand and customer behaviours. This knowledge was essential for optimizing the shared mobility platform and increasing its profitability.

The main questions we wanted to answer were:

• How the number of rentals changes throughout the day?
• Do local events influence demand?
• Which routes are the most popular?
• Where customers rent and return vehicles?
• Are there any areas where vehicles are idle for a long time?
• Who are the best customers?
• Which customers are about to leave and need the attention?
• Do customers have their regular routes?

The insights generated through customer behaviour analysis have a direct impact on business performance and can be instantly used for the benefit of the business. For example, knowing the locations where users usually rent and return vehicles helps the provider optimize the positioning of vehicles and thus – satisfy customer demand better and maximize the utilization of vehicles.

To get a wide knowledge of the shared mobility platform and its users we made a number of hypotheses and applied various data analysis techniques (statistical analysis, RFM analysis, cohort analysis, geospatial data analysis) to check them.

The inherent part of any data analysis (especially geospatial data) is visualization. To make the results of our analysis understandable and easily accessible we used various data visualization tools, including choropleth maps, heat maps, point maps and cluster maps to present geospatial data in a comprehensible way.