IRJET- Road Traffic Prediction using Machine Learning

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 04 | Apr 2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 498
Road Traffic Prediction using Machine Learning
Bhaskar Chowdhury1, Maithili Kinhikar2, Mrs. N. Noor Alleema3
1,2Student, SRM Institute of Science and Technology, Chennai, India
3Assistant Professor, Dept. of Information Technology, SRM Institute of Science And Technology, Tamil Nadu, India
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - There are numerousinventoriesincarbusinesses
to structure and assemble wellbeing measures for vehicles,
however car crashes are unavoidable. There is a colossal
number of mishaps winning in all urban and rustic regions.
Examples required with various conditions can be recognized
by building up a precise expectation models which will be fit
for programmed detachment of different inadvertent
situations. These group will be valuable to forestall mishaps
and create wellbeing measures. This idea accepts to secure
greatest conceivable outcomes of mishap decrease utilizing
low spending assets by utilizing some logical measures.
Key Words: Safety measures, accurate prediction,
forestall, Mishaps, conceivable outcomes, wellbeing
1. INTRODUCTION
There is a colossal effect on the general public becauseofcar
crashes where there is an incredible expenses of fatalities
and wounds. In current years, there is a significantriseinthe
researches and study of mishaps that has determines the
intensity of the injuries caused to the road accidents.
Detailed and informative data on the nature of accidents
provides a way for improved and accurate predictions. The
right and effective use of accident records lies on some
factors, like the authenticity and integrity of the data, record
retention, and data analysis. There is numerous
methodologies connected to thissituationtothink aboutthis
issue. An ongoing report showed that the private and
shopping locales are more perilous than town areas.as may
have been anticipated, the frequencies of the setbacks were
higher close to the zones of habitation perhaps due to the
higher presentation. A survey uncovered the fact that the
mishap rates among the dwelling areas are categorized as
relatively underprivileged and noticeably higher than those
from relatively prosperous areas. Thus, it is step to curb
down the accidents to certain extent.
2. LITERATURE SURVEY
This project is a special, important and overall leaves no
stones unturned. With known priorities that mankind has
been upto lately are major focuses on materialistic things
and the advent of corporate world worsening the situation.
For a whole in one, the project gives you what needs to be
given to world right now. The project deals with four pillars
of Information Technology, Machine Learning, Application
development, pipelining and Bridging the gap between
Technology and Science. We deal with immense pressure
when we put out the best of Technology and blendtogoodof
society and it is what the project is all about. In regards to
the project addresses the public and private levels thus
attacking the root and trunk all at the same time. Digging
deep into the work of some researchers here, a deep
understanding of the analysis could be made. Mr. Sachin
Kumar used data mining techniques to find the locations
where accidents are of high rate and then analyze them to
identify the factors that have an effect on road accidents at
that locations. The dataset has been divided into different
categories based on the types of accidents occurred using k-
means algorithm. At that point, affiliation rule mining
calculation connected so as to discover the connection
between particular properties which are in mishap
informational index and as indicated by that know the
attributes of areas. Mrs. S. Shanthi classified dataseton basis
of gender using RndTree to get accurate results. From the
Critical Analysis Reporting Environment (CARE) system
provided by the Fatal Analysis Reporting System (FARS)
used by the training data set. Also Tessa K. Anderson
introduced anidea ofidentifyinghigh-densityaccidentprone
areas, which creates a clustering technique that indices can
be compared in time and space. The estimation tool enables
the better view and handling of density-based events, which
helps to identify the accident prone areas. The rate of
damage occurring during a trafficaccidentissimulatedusing
the caliber of various machine learning paradigms, such as
neural networks trained using certain learning method,
decision trees, and concurrent mixed models involving
neural networks and decisiontree.Afterthe experiments the
results show that the hybrid decision tree neural network
method is better than the singlemethodinmachinelearning.
3. REQUIREMENTS
Software: Anaconda – Jupyter.
Language: Python3
Modules Used:
 numpy
 pandas
 From pandas. tools. plotting import scatter_
matrix

 import matplotlib. pyplot as plt
 windows 7 or above
 RAM 2GB or more
 Processor i5 5th Gen or above
4. EXISTING SYSTEM
Many research examines concentrated exclusively on
recognizing the major factors that reason street crashes.
From these examinations, it was seen that human variables
have the huge effect on mishap chance. The fundamental
components impact on street security straight forwardly
identified with the driver are i.e., driving conduct, driver's
impression of traffic dangersanddrivingknowledge.Drivers
include as often as possible in demeanors that reason street
wellbeing issues. A large number of theseframesofmind are
dynamic, cognizant standard infringement, while others are
the consequence of blunders because of less driving
knowledge, fleeting slip-ups, negligence or inability to
perform work, the last frequently identified with age. These
practices frequently add to car accidents. Other than of
unsafe driver conduct the terrible drivingpracticesandpoor
information alongside lack of respect for street and
wellbeing guidelines are the conspicuous issues.
Disadvantage:
1. The examination researched that the assignment of
driving can be simple or troublesome relying upon
the passing errand request of driving and the
driver's ability to control his/hervehicleeffectively.
2. The examination explored that the task of driving
can be basic or troublesome depending upon the
passing errand solicitation of driving and the
driver's capacity to control his/her vehicle
successfully.
5. PROPOSED SYSTEM
Models are madeutilizing mishapinformation recordswhich
can comprehend the qualities of numerous highlights like
driver's conduct,roadwayconditions,lightcondition,climate
conditions, etc., which can in turn help userstofigureoutthe
safety methods which will be beneficial to avoid mishaps. It
very well may be outlined how factual technique dependent
on coordinated charts, by contrasting two situations
dependent on out-of-test gauges. The model is performed to
recognize measurably noteworthy components which can
almost certainly anticipate the probabilitiesofaccidents and
damage that can be utilized to play out a hazard factor and
decrease it. Here the road mishap study is done by analyzing
certain information by giving some queries which is related
to this study. The queries are about the most dangerous
time to drive, percentage of accidents occurring in rural,
urban and other areas, trend in the number of accidentsthat
occur every year, accidents in high speed limit areas have
more casualties or not and so on. These data can beacquired
using Microsoft excel sheet and the required answer can be
found. This analysis aims to emphasize the data that is
valuable in a road traffic accident and allow estimates to be
made.
ADVANTAGES:
1. Very detailed analysis is done based on
various factors which helps in getting
accurate results.
2. It covers a wide topological range.
6. METHODOLOGY
6.1 ACCIDENT DATASET
The dataset has been divided into five categories to classify
the intensity of accidents. This classification gives a better
understanding of the datasets.
6.2 DATA PREPARATION
This represents the manner in which the collisions have
occurred ,which are divided into seven categoriessuchasno
collision , rear to rear, head-on, head-rear, side-wipe same
direction , side-wipe opposite, angle collision. The data has
been prepared in such a manner to provide a broader view
on collisions.
6.3 STATISTICAL ANALYSIS
The earlier analysis only focused on two categories which
were injury, non-injury. This research has been further
extended to introduce few more categories which are
possible injury, non-incapacitating, incapacitating and fatal
injury.
6.4 PERFORMANCE COMPARISON
Since more categories have been added in the analysis so
this has helped to hike the performance of the proposed
system.
6.5 RESULTS
 Accident prediction in urban, rural and other areas
Fig 6.5.1

The above graph shows what fraction ofaccidentoccurred in
different areas, which are urban, rural and others
 The trend in the number of accidents that occur
each year
6.5.2
The above graph gives the correlation between numbers of
accidents in the years2005-2015.Thenumberofaccidentsin
the y-axis are taken from the sampledatasetintothetraining
dataset for annual predictions.
7. CONCLUSION
A more broad analysis of the road accident can be made
which can help improve the predictions, making them more
accurate. Also this is a very efficient way than the earlier
approach which did not cover wide range of factors making
it less effective in moderndayaccidentpredictions. Wemade
the analysis even wider, keeping in mind the ever surging
traffic accidents and with the aim of curbing the inevitable
accidents to some extent.
REFERENCES
[1] Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, “Electron
spectroscopy studies on magneto-optical media and
plastic substrate interface,” IEEE Transl. J. Magn. Japan,
vol. 2, pp. 740–741, August 1987 [Digests 9th Annual
Conf. Magnetics Japan, p. 301, 1982].
[2] M. Young, The Technical Writer’s Handbook. Mill Valley,
CA: University Science, 1989.
[3] Sachin Kumar, Durga Toshniwal, "A data mining
approach to characterize road accident locations”, J.
Mod. Transport. (2016) 24(1):62–72.
[4] S. Shanthi and Dr. R. Geetha Ramani, “Gender Specific
Classification of Road Accident Patterns through Data
Mining Techniques”, IEEE-International Conference on
Advances in Engineering, Science and Management
(ICAESM -2012) March 30, 31, 2012.
[5] Tessa K. Anderson, “Kernel density estimation and K-
means clustering to profile road accident hotspots”,
Accident Analysis and Prevention 41 (2009) 359–364.

IRJET- Road Traffic Prediction using Machine Learning

More Related Content

What's hot (20)

Similar to IRJET- Road Traffic Prediction using Machine Learning (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Road Traffic Prediction using Machine Learning