Road Accident Prediction and Analysis Using Machine Learning
Road Accident Prediction and Analysis Using Machine Learning
Volume 8, Issue 1, January – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Road Accident Prediction and Analysis Using
Machine Learning
Ms.Preshita Bhortake, Vivek Barhate
Department of Computer Engineering, Marathwada Mitra Mandal’s College of Engineering,
Savitribai Phule PuneUniversity of Pune, Pune, India
Abstract:- India has a substantial population, which II. ABBREVIATIONS
contributes to a significant daily automobile commuter
population. This leads to several accidents occurring KNN- K-Nearest Neighbor
every day. These mishaps frequently result in severe LR-Logistic Regression
financial hardship for families as well as the possibility of 3.GNB-Gaussian Naïve Bayes algorithm
fatalities. The goal of this article is to identify accident- 4.VR-Virtual Machine
prone areas and alert regular commuters to the incidents
that are occurring there. Accidents can occur at any time III. METHODOLOGY
and without warning, but as users of this interface, we
can be more cautious in locations where accidents occur The Virtual Machine (VR): It has the trained and tested
frequently. The user interface will alert a user to the Machine learning algorithm implemented. The frontend
high-medium accident risk areas. and backend server are deployed on it.
Front-End: Users input for the prediction factors are
Keywords:- Random Forest Algorithm,GaussianNaïve Bayes taken and sent to the backend server.
algorithm, Logistic regression, Machine Learning. Back-End: The model is deployed here and the input data
is fed into the Machine Learning model.
I. INTRODUCTION Machine Learning Model: We have used decision tree,
random forest and logistic regression and also applied
According to the study conducted by WHO, millions
hyperparameter tuning to increase its efficiency.
of people die as a result of traffic accidents every year.
Accidents are the primary problems facing the world today Modules Used:
since they frequently result in numerous injuries, fatalities, Data Collection: Collection of data from repository
and financial losses. Road accidents are a problem that has dataset is the initial step for the commencement of the
impacted the general public's well-being and the economy of research work in which data is collected from various
the country.A fundamental task for road transportation previously used dataset andpatient history.
systems is to develop precise models to identify the cause of Attribute Selection: Attributes are properties of the
accidents and provide recommendations for safe driving. dataset which are used for determining factorslike
This research effort creates models based on the factors that driver behaviour, road condition etc. It is a
cause accidents, such as weather, causes, road cumbersome process to select the attribute as
characteristics, road conditions, and accident type. Likewise, itsselection affects the accuracy and continuousness of
select a number of significant elements from the best model the training model.
in order to build a model for describing the cause of
Data Pre-processing: Transforming data into required
accidents. In more detail, these are algorithms like Random
format and adjusting the missing dataset values,
Forest Algorithm, Gaussian Naïve Bayes algorithm, Logistic
attributes scaling etc. It adds or alter the dataset
Regression (LR) implemented to discover the impact of
various factors and gives a safe driving protocol. Key Management: This module is not only
responsible for the generation and maintenance of the
key, but also uses the hierarchical key management
approach to protect the key.
Balancing of Data: Imbalanced dataset are balanced
1. Under Sampling- Reduction of dataset; When ample
amount of dataset is there.
2. Over Sampling- Increase in dataset; When Dataset
isless.
IJISRT23JAN141 www.ijisrt.com 1018
Volume 8, Issue 1, January – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Fig 1: Flowchart Diagram
IV. ALGORITHMS trees based on data samples, obtains predictions from each
one, and then votes to determine which is the best answer.
A. Random Forest algorithm Because it averages the results, the proposed technique is
It is a supervised learning technique that is utilized for both superior to a decision tree classifier excessively.
classification and regression.However, classification issues are
its primary usage. Random forest basically builds decision
Fig 2: Random Forest
B. Decision Trees of any other feature. The NaïveBayes algorithm's key benefit
In decision analysis, a decision tree can be used to represent is that it is an easy-to-use yet effective method.
decisions and decision making visually and explicitly. As the
name goes, it uses a tree-like model of decisions. Though a D. Decision Tree Algorithm
commonly used tool in data mining for deriving a strategy to It is a subset of the supervised learning algorithm family.
reach a particular goal, it’s also widely used inmachine Unlike the other supervised learning techniques, the decision
learning. tree methods can be implemented to both classification and
regression issues. The main goal of employing decision trees
C. Gaussian Naïve Bayes algorithm is to build a training model that can be applied to forecast the
It is the Naïve Bayes extension. A probabilistic machine type or value of a target variable by learning choice rules
learning approach called Naive Bayes can be applied to a inferred from historical data. The decision tree approach uses a
variety of categorization applications. Naive Bayes is tree representation to attempt to address the issue. Each leaf
frequently used for document classification, spam filtering, node of the tree corresponds to a class label, whereas each
prediction, and other tasks. This method takes its name from internal node of the tree relates to an attribute.
Thomas Bayes' discoveries, on which it is based.The method
includes features in its model that are independent of one
another, hence the term "Naïve." Any changes to the value of
one feature of the algorithm have no direct effect on the value
IJISRT23JAN141 www.ijisrt.com 1019
Volume 8, Issue 1, January – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
V. CONCLUSION
The training of a model that can detect multiple
parameters to predict and assess road accidents can be done
using machine learning techniques. It defined the requirements
for each technique and developed a comprehensive framework
to identify and stop road accidents. It is evident that a variety
of elements, including vehicle kinds, vehicle ages, weather
conditions, road conditions, and others, have a significant
impact on incidents of traffic accidents. Creating an
application that accurately predicts traffic accidents based on
the aforementioned parameters is what future need is.
ACKNOWLEDGEMENT
I take this to express my deep sense of gratitude towards
my esteemed guide Prof. J. A.Wagh for giving me this
splendid opportunity to select and present this seminar and
also providingfacilities for successful completion.I thank Dr.
K. S. Thakre, Head, Department of Computer Engineering, for
opening the doors of the department towards the realization of
the seminar, all the staff members, for their indispensable
support, priceless suggestion and for most valuable time lent
as and when required. With respect and gratitude, I would like
to thank all the people, who have helped me directly or
indirectly.
REFERENCES
[1.] M. F. Labib, A. S. Rifat, M. M. Hossain, A. K. Das and
F. Nawrine, "Road Accident Analysis and Prediction of
Accident Severity by Using Machine Learning in
Bangladesh," 2019 7th International Conference on
Smart Computing & Communications (ICSCC), 2019,
pp. 1-5, doi: 10.1109/ICSCC.2019.8843640
[2.] J. Patil, V. Patil, D. Walavalkar and V. B. Lobo, "Road
Accident Analysis and Hotspot Prediction using
Clustering," 2021 6th International Conference on
Communication and Electronics Systems (ICCES),
2021, pp. 763-768, doi:
10.1109/ICCES51350.2021.9489074. Ding Chen,
Qiseng Yan, Chunwang Wu, Jun Zhao “SQL Injection
Attack Detection and Prevention Techniques Using
Deep Learning”, School of Cyberspace Security,
Chengdu University of Information Technology,
Chengdu, 2020.
[3.] Priyanka A. Nandurge, Nagaraj V.
Dharwadkar,Analyzing Road Accident Data using
Machine LearningParadigms, International conference
on I-SMAC (IoT inSocial, Mobile, Analytics and
Cloud), 2017, pgno.604-610, IEEE.
[4.] S. Ahmed, M. A. Hossain, M. M. I. Bhuiyan and S. K.
Ray,” A Comparative Study of Machine Learning
Algorithms to Predict Road Accident Severity,”, Proc.
IEEE Int. Conf, pp.1464-1473, Dec. 2020.
[5.] Dr. Shivashankara S, Archana B J, Vijayalakshmi D,
Meghana Gowda D, Anusha “Road Accident Analysis
and Prediction Using Data Mining.
