Descriptive and predictive models
Open Source communities & application developers
Public authorities, Urban planners & Citizens
Toolbox of descriptive and predictive models
High-level services, implemented as LEMONADE modules, are Spark and COMPSs building-blocks for the final-user applications addressing a set of high value services. These services are exploited by the final user applications from the project, but are generic enough to be of interest for external application developers. These services are:
- Traffic Congestion Prediction, which aims to identify traffic jams using data provided by Waze, an application widely used by drivers to obtain trajectories to destination or notifications regarding unusual traffic behavior, such as traffic jams, accidents or closed roads. For that end, this service uses a probabilistic graphical model equipped with Gaussian latent nodes. (https://github.com/eubr-bigsea/waze-jams).
- Trip Duration Prediction is a tool that aims to predict bus trips duration based on historical bus GPS data. We train the model using state-of-the-art Machine Learning techniques on historical bus trips data, and use it to predict future trips. (https://github.com/eubr-bigsea/btr-spark).
- Sentiment Analysis is a service that transforms social media data (textual) into a quantitative estimation of the citizens expressed sentiment. Such analysis targets a specific subject, for example, traffic situation or city services, or a population of a region. (https://github.com/eubr-bigsea/Lemonade_apps/tree/master/sentiment_analysis)
- Trip Crowdedness Prediction is a tool that aims to predict the number of passengers (crowdedness) of a bus trip in the future, based on historical bus location and ticketing data. (https://github.com/eubr-bigsea/btr-spark).
- People Paths is a service application that performs a descriptive analysis on bus GPS and passenger ticketing data, finding paths taken by Public Transportation city users in a time period, and matching the paths origin/destination locations with city area social data: population, income and literacy rate. (https://github.com/eubr-bigsea/people-paths).