This challenge is organized as part of the recruitment, in early 2019, of 2 to 3 trainees in the Paris fire service seeking to work in in data analysis, data science and / or business intelligence. Any other participant is however welcome.

The challenge concerns an operational problem of the rescue service of the Paris Fire Brigade. The anonymised data of the service are made available.

Candidates must conduct in a Jupyter Notebook the analysis of the dataset and answer the problem even extend it. We will pay close attention to the graphical transcription of data, comments and conclusions. A geographical representation is unavoidable.

  • Challenge opening date: Friday 14 December 2018
  • Challenge closing date: Sunday 20 January 2019

Retained candidate

Geoffrey Chetcuti
Student in the Institut supérieur d’électronique de Paris / ISEP.
Business Intelligence Specialization

ISEP is a French Grande école located in Paris. It specializes in electronics, telecommunication and computer science.
Specialization details


Problem

Some road sections and periods of traffic in Paris, the departments of 92, 93 and 94 are problematic during the delivery of emergency relief, will you be able to identify them?

The vehicle movements we are interested in for this analysis are those that occurred during the transit to a place of emergency request, that is to say movements between the Departed and Presented positions. (see Operational Status)

Note that some emergency vehicles do not transmit their GPS positions.

Other optional issues
  • Estimate the catchment areas 10 minutes from the rescue centers, for different time slots of the day (you do not need the location of the rescue centers to evaluate these areas).
  • The current dispatch system uses a model of estimation of the transit times for emergency vehicles very rigid, would you be able to elaborate a more relevant estimate using for example: the time of the day, day of the week or day of the weekend, road traffic mearsures or weather conditions.
  • Are you able to extend the spectrum of your analysis leading to findings that managers of a rescue service might be interested in.

Tutorials

A consultation of the tutorials is highly recommended. They are here to help you.

Works submission

  • The works will necessarily be restored within one or more Jupyter Notebook (file .ipynb) in Python 3 preferably, however R will be tolerated.
  • Must be sent no later than the 20th of January 2019 by email to benjamin.berhault@pompiersparis.fr
  • We will pay close attention to the graphical transcription of data and conclusions.
  • Geographical rendering is strongly recommended.

Data


The data cover a period from Monday, September 5, 2016 to Sunday, October 30, 2016 inclusive, or 8 weeks.

Dataset columns
  • Date - Date time of the observed value
  • Longitude - Longitude of the emergency vehicle GPS position
  • Latitude - Latitude of the emergency vehicle GPS position
  • Véhicule de secours - Emergency Vehicle Identifier
  • Famille de véhicule de secours - Family of emergency vehicle
  • Statut opérationnel - Operational status of the emergency vehicle, manually entered by crew
  • Numéro intervention - Identifier of the rescue request to which the emergency vehicle has been allocated
  • Type intervention - Color referring to an emergency request category
  • Availability - Emergency vehicle availability based on status (0: unavailable, 1: available)
  • Centre de secours - Identifier of the fire station from which the emergency vehicle came
Operational statutes

A emergency vehicle transmits various radio statutes during an intervention for a rescue request, within the dataset these have been reduced to the essential information:

  • Sélection - selection of rescue vehicle in the dispatch application
  • Parti (Departed) - the vehicle begins its route leading to emergency request place
  • Présenté (Presented) - the vehicle arrives at the place of the emergency request
  • Transport hôpital - the vehicle starts transporting a victim to a hospital
  • Arrivée hôpital - the vehicle arrives at the hospital
  • Quitte hôpital - the vehicle leaves the hospital
  • Rentré - the vehicle has returned positioned at its parking spot (a fire station)
  • Quitte les lieux - because the vehicle can also simply leave the scene of an intervention without having to carry a victim
  • Indisponible - for various reasons the vehicle can be put in the unavailable position
  • Non pertinent - statutes without interest for that study

External sources of potentially relevant data