About Project
Introduction
The rise in the number of vehicles is bound to have some adverse effects, including, but not restricted to, traffic congestion in densely populated areas and overburdening of the infrastructure in an urban area. It also causes a higher number of deaths yearly, related to traffic.
The latest research undertaken by the World Health Organization (WHO) shows that approximately 1.25 million individuals die each year as a result of road accidents and millions more are injured, almost 49% of which are passing pedestrians and cyclists. If certain precautionary measures are not taken to overcome this percentage, then, by the year 2030, traffic accidents will become the fifth major cause of deaths. In this respect, instruments and systematic methods to improve road safety have been developed in the scientific community for many years.
The following points may prove invaluable in reducing the number of casualties and injuries drastically, caused by road accidents:
Providing medical help to victims of road accidents in a timely manner.
Notification about the precise situation to the first responders.
Lack of a permanent database, holding all appropriate documents and records, which can be examined as and when necessary.
When it comes to emergency response to road accidents, every second counts. With heavy traffic patterns and the unique layout of the city, finding the best locations to position emergency responders throughout the day as they wait to be called is critical in a country like India.
How Data Analytics can help?
Data Analytics for this problem can be helpful in different ways:
could improve the emergency response
decrease the distance between ambulances and crashes by predicting the future possibility of crashes in locations
increase our understanding of the reasons for crashes
Datasets
For this task, we are using the training data (recorded crashes up to June 2019) along with supplementary data from Uber Movement, road survey data, and weather patterns to identify patterns of risk across the city. Then we are using these findings to place six-seven virtual ambulances around the city, moving them around throughout the day with the goal of minimizing the distance traveled when responding to crashes during the test period.
contains crashes between 2018-01-01 and 2019-06-01, each of which has a location (latitude and longitude) and a time.
data from road segment surveys
has daily weather of the city
Approach
Data Preprocessing - To clean the dataset and modify it to meet our project requirements.
Data Visualization - Visualizing the data so that we can make the most information out of it.
Clusterization - By clustering the points of latitudes and longitudes of crash sites according to the time, distance. We can also find the area nearby frequent crash sites which can be optimal to place the ambulance.
Multiple Linear Regression(MLR) - for predicting latitudes and longitudes of future crashes and then clusterization. This can be done to predict what can be the possible crash sites in the future and based on that we form clusters and find the optimum 6-7 points where we can position the ambulances.