Solicitarinformación

Este sitio Web utiliza cookies propias y de terceros con objeto de mejorar la experiencia de navegación. Si continúa navegando estará aceptando de forma expresa el uso de estas cookies. Puede obtener más información en nuestra página Uso de Cookies

BIDEMAP. A project led by Dinycon Sistemas with the participation of Tecnalia for the development of predictive models.

BIDEMAP: Behavioural analysis of urban pedestrian mobility

The main objective of the BIDEMAP project is the development of a tool to support municipal managers in the behavioural analysis of aggregate pedestrian mobility in the Smart City: occupation of public spaces, mobility flows, Origin-Destination matrices, heat maps, etc.

To build these high value-added services for the city from the mobility data collected by Dinycon, it is necessary to build data analysis solutions to detect patterns, trends, describe behaviours and predict them, and translate them into attractive and interpretable visual formats for the city manager. In short: to convert the massive, dry and unmanageable digital data into enriched and interpretable information for the human operator.

To build the analytical layer of the data collected, Dinycon relies on Tecnalia and its long experience in Smart Mobility Lab, a behavioural mobility analysis laboratory based on the Big Data paradigm.

The combination of Dinycon experience in the acquisition of quality data and Tecnalia experience in the analysis of massive volumes of data in the field of mobility has encouraged us to present this project to the HAZITEK programme.

From the historically recorded data, short- and long-term predictive models are constructed, and correlations between sensors are established.

SHORT-TERM FORECAST

A predictive model is built that has as input a window of readings prior to the data to be predicted, and as output the predicted data. Two variables are defined: prediction depth and prediction horizon.

LONG-TERM FORECAST

The long-term forecasting method is based on the detection of patterns in the known data. The data are grouped by days, and these are grouped by similarity, forming patterns. In this way, for example, all Saturdays, which have a similar flow profile, can be grouped together.

CORRELATION BETWEEN SENSORS

Analogous to short-term prediction. Contextual prediction establishes correlations with other sensors, establishing a relationship between them even if they do not depend on each other. This can be seen in the following drawing

HEAT MAPS

For the visualisation of the flows recorded by the sensors, we chose an interactive graph that allows us to observe the traffic of people at each instant of time. The objective is to check if there are correlations between the exits and entrances of each sensor. That is to say, if when a lot of people enter through a sensor, they also leave; they could be considered points of constant agglomeration; also to analyse if the increases in the values registered by a sensor are matched by the increase in the values of another sensor or, on the contrary, their decrease.

BIDEMAP is funded by the HAZITEK 2019 Business R&D Support Programme.