CHAPTER 1

INTRODUCTION

1.1 OVERVIEW
Road Traffic Accident (RTA) is an unexpected event that unintentionally occurs on the
road which involves vehicle and/or other road users that causes casualty or loss of property.
Over 90% of the world’s fatalities on roads occur in low and middle income countries which
account for only 48% of world’s registered vehicles. The financial loss, which is about
US$518 billion, is more than the development assistance allocated for these countries. While
developed rich nations have stable or declining road traffic death rates through coordinated
correcting efforts from various sectors, developing countries are still losing 13% of their
gross national product (GNP) due to the endemic of traffic casualties. World Health
Organization (WHO) fears, unless immediate action is taken, road crashes will rise to the
fifth leading cause of death by 2030, resulting in an estimated 2.4 million fatalities per year.

Accidental severity analysis involves three factors that are number of injuries, number of
casualties, and destruction of property. Authors take severity level independently and
consider four options like light injury, severe injury, fatal, and property damage. Road traffic
accidents are a major cause of injuries, deaths,permanent disabilities and property loss. It not
only affects the economy it also affects the healthcare system because it puts a burden on the
hospitals. Statistics shown by the ministry of public security of china from the years 2009
and2011, traffic accidents caused an average of 65123 people to lose their life and 255540 got
injuries annually. Identifi-cation of primary factors affecting road accident severity is required
to minimize the level of accidental severity. Acci-dental Severity does not happen by chance;
there are patterns that can be predicted and prevented. Accidental events can be analyzed and
avoided. Being one of the major issues of accident management, accident severity prediction
plays an important role to the rescuers in evaluating the level of severity in traffic accidents,
their potential impact, and in implementing efficient accident management
procedures.Accidental severity analysis involves three factors that are number of injuries,

1
number of casualties, and destruction of property. Authors take severity level independently
and consider four options like light injury, severe injury, fatal, and property damage. The
accidental severity level is definedas injury, possible injury, and property damage .

The fast-growing population, increasing motorization, and rapid urbanization have made
road users more vulnerable to accidents which are on the rise globally and are a significant
challenge to deal with. Road traffic accidents impose a substantial human, financial, and
economic burden on society. According to the report published by the World Health
Organization (WHO), road crashes result in an average annual human death of approximately
1.3 million and 20 to 50 million serious and minor injuries. Countries and international
organizations have designed technologies, systems, and policies to prevent accidents. Whilst
several measures have been adapted to make the roads safe for users, there is no such
real-time warning system available to guide the user about the probability of an accident.
Similar to multiple other countries in the world, New Zealand also has road signs, such as
high crash areas, slippery when wet, speed limits, etc., which warns users regarding road
safety. Apart from the road signs, prediction of accident rates on road links is an important
input to road safety improvement. The prediction model explores the relationships between
crash severity injury categories and contributing factors such as driver behavior, vehicle
characteristics, roadway geometry, road-environment conditions and cause of accidents,
which could then be addressed by transportation policies. Vehicular crash data can be used to
model both the frequency of crash occurrence and the degree of crash severity. Crash
frequency refers to the prediction of the number of crashes that may occur on a specific road
segment or intersection over a certain time frame.

The accidental severity level is defined as injury, possible injury, and property damage.
In the last two decades, accidental severity is one of the popular research areas. Researchers
were applying different statistical approaches for road accident classification. These
techniques help in analyzing the cause of road accidents. To uncover the underlying
relationship between a road accident and the contributing factors, ML based models have
been studied in recent days. The primary focus of this study is to analyze a set of widely used

2
ML models in terms of their prediction accuracy with the variation of contributing factors.
Therefore, this research aims to study different ML algorithms and compare the performance
of these algorithms which can be considered to predict road accidents and its severity.

1.2 ENVIRONMENT FACTORS CAUSING ACCIDENT

Environmental factors causing accidents is a term used to capture all the causes of a
workplace incident that can be directly attributed to the working environment. These include
features of the natural environment, aspects of workplace design, as well as the machinery or
equipment used on the jobsite. Some of the environmental factors that are contributing to the
accident are listed below.

i. Road Surface

Road surface or pavement distress is a condition of road damage such as cracking,

patching and potholes, surface deformation, and surface deflect. This project discusses the
effect of pavement distress on the risk of accidents based on collective damage conditions of
pavement surfaces. Road surface condition is a matter of concern especially for traffic safety.
Poor road surface conditions (large potholes and deep cracks exist, discomfort at slow speeds)
can lead to high rates of accidental fatality, especially in high speed roads in single vehicle
and multi vehicle accident types.

Fig 1.1 Number of accident in different road type

3
The figure 1.1 mimics the number of accidents that happen on different types of road surfaces
in Europe. The A, A(M), B and the motorway are the roads that are maintained by the
highway authorities. The unclassified are the soil roads in the rural and metro areas without
the maintenance, which kinds of roads led to the most of the accidents.

ii. Lighting Condition

Inadequate lighting can lead to serious injuries resulting from slips and falls, collisions
or other similar incidents which can break bones, cause serious ligament damage, concuss
victims or even threaten their lives. In addition to headaches, your workers may also
experience neck, back and shoulder pain if they have to hunch over and strain to see items as
a result of incorrect lighting design.

Fig 1.2 Number of accident on different lighting conditions

4
The figure 1.2 most of the accidents happen in the daylight but when street lights are present
it is in the middle day-night evening times .

iii. Weather Condition

Weather acts through visibility impairments, precipitation, high winds, and temperature
extremes to affect driver capabilities, vehicle performance (i.e., traction, stability and
maneuverability), pavement friction, roadway infrastructure, crash risk, traffic flow, and
agency productivity.

Weather conditions can play a part in many accidents, even those concerning other
drivers who may have caused your accident and resultant injuries. Sometimes, it’s not until
you’re involved in an accident that you realize just how dangerous it can be to drive in
adverse weather conditions.

Black Ice : Black ice is caused by rain or snow melting and freezing on a road or footpath. It
can also form due to a road not being designed or maintained well enough to drain water
effectively. It’s hard to see and even harder to drive on. As a result, it plays a leading role in
many weather-related accidents.

Rain : Rain can affect the ability to drive safely in many ways. It might impact driver
visibility, causing you to be caught up in an accident that was out of your control. It can also
cause roads to be slippery and even flood when road drainage systems are poorly designed or
maintained. If you must travel in the rain, pay attention to the weather forecast of the areas
you’re traveling through and drive to the conditions.

Wind : Driving in high winds can be challenging for many people. The windier it is, the more
hazards there may be on the road, and the harder your vehicle may be to control. High-sided
vehicles like trucks and vans may be more at risk than others, alongside motorcycles and
inexperienced operators. Wind can also cause debris to fly onto roads, directly into the path of
fast-moving traffic. This can also lead to expensive and painful accidents.

5
Snow : While many people have mastered driving in the snow in the area they live in, it
doesn’t mean accidents don’t happen. Snow can act unpredictably while also hiding layers of
treacherous ice that can lead to spin outs and collisions. City officials may arrange for roads
to be regularly maintained for safety, but it’s generally advised for people to avoid driving in
such weather conditions if they don’t need to.

Sun : Many people use sun blindness as an excuse for causing or being involved in a car
accident. Sun blindness involves poor visibility due to the sun being directly in your eyes
while driving.

Fig 1.3 Number of accident on different weather conditions

The weather can wreak havoc on the roads, and accidents happen. These weather events
above may be some of many contributing factors to an accident you or a loved one were in,
and you may wish to discuss them with your chosen personal injury attorney. Weather
conditions can play a part in many accidents, even those concerning other drivers who may

6
have caused your accident and resultant injuries. Sometimes, it’s not until you’re involved in
an accident that you realize just how dangerous it can be to drive in adverse weather
conditions.

The general objective was to disaggregate the road accident problem using mass
accident data to find groups of road users, vehicles and road segments which would be
suitable targets for countermeasures. Successful development of a countermeasure requires a
clear understanding of where it can potentially break the chain of events leading to traumatic
injury on the road. A countermeasure is a measure which attempts to break the road trauma
chain before one of the undesirable steps can occur (eg. accident involvement, injury or
death). A target group for a countermeasure is a group of entities (humans, vehicles or roads)
for which the chain can be broken effectively and, desirably, cost-effectively.

Mass accident data needs to be analyzed to find target groups for countermeasures in a
way which maximizes the chances that the countermeasure will be cost-effective. The project
has developed general principles for analysis which meet this aim. These have led to four
specific methods of mass data analysis, depending on the nature of the road trauma problem
being addressed in the search for countermeasure target groups, namely:

● High Risk Groups (groups with high rates of accident involvement per opportunity to
be involved)
● High Severity Groups (groups with high rates of severe injury per accident
involvement)
● Accident Involvement Clusters (groups involved in accidents who are homogeneous
on a number of factors relevant to countermeasure design and as large as possible)
● Severe Injury Clusters (groups associated with severe injury who are homogeneous
on a number of factors relevant to injury countermeasure design and as large as
possible).

7
1.3 IMPACTS OF ROAD ACCIDENT

Road traffic accidents (RTA) are a global problem resulting in deaths, physical injuries,
psychological problems and financial losses. These financial damages have immediate
consequences and long term consequences on the victims and their families. Road traffic
injuries are the leading cause of death globally and the principal cause of death in the age
group of 15 to 49 years. Every year the lives of approximately 1.3 million people are cut short
globally as a result of a road traffic crash.

i. Social Impacts

The social consequences of road traffic accidents include loss of productivity of the
victims, the cost of the legal system, the cost of pain and suffering and loss of quality of life
of the victim and their family. The loss of productivity represents a significant proportion of
the total social costs.

ii. Economical Impacts

In economic terms, the cost of road crash injuries is estimated at roughly 1% of the
gross national product in low‐income countries, 1.5% in middle‐income countries and 2% in
high‐income countries.

1.4 OBJECTIVE

There are many factors that contribute to road accidents, especially those related to the
environment, vehicles and the travelers. Despite many precautionary measures to reduce road
accidents, it remains one of the uncontrollable. Many factors are associated with traffic
accidents, some of which are more significant than others in determining the severity of
accidents. Identifying the accident severity and its factors can be given a solution for
reducing the risk of future road accidents. Successful development of a countermeasure
requires a clear understanding of where it can potentially break the chain of events leading to

8
traumatic injury on the road. The use of big traffic data and artificial intelligence may help
develop a promising solution to predict or reduce the risk of road accidents. Machine
learning techniques are playing a vital role in predicting the future events based on the
historical data. Developing the machine learning classifier for the road accident severity
prediction can help to identify the accident severity and developing countermeasures and
alerting systems for the future preventions.

To uncover the underlying relationship between a road accident and the contributing
factors, ML based models have been studied in recent days. The primary focus of this project
is to analyze a set of widely used ML models in terms of their prediction accuracy with the
variation of contributing factors. Therefore, this research aims to study different ML
algorithms and compare the performance of these algorithms which can be considered to
predict road accidents and its severity.

The project’s major purpose is to:

● Identify the group of factors lead to the road accident

● Develop a system to identify the road accident severity
● Deploying a model with higher accuracy and lower error rate
● Developing countermeasures for accident injuries
● Developing E-mail based road accident tracing system
● Developing an intelligent AI based application which can notify the user about their
zonal accident severity based on their live location environmental factor

9
 CHAPTER 2
LITERATURE SURVEY

1. TRAFFIC ACCIDENT’S SEVERITY PREDICTION: A DEEP-LEARNING

APPROACH-BASED CNN NETWORK

In a traffic accident, an accurate and timely severity prediction method is necessary for
the successful deployment of an intelligent transportation system to provide corresponding
levels of medical aid and transportation in a timely manner. Existing traffic accident’s severity
prediction methods mainly use shallow severity prediction models and statistical models. To
promote the prediction accuracy, a novel Traffic accident’s severity Prediction-Convolutional
Neural Network (TASP-CNN) model for traffic accident’s severity prediction is proposed that
considers combination relationships among traffic accident’s features. Based on the weights
of traffic accident’s features, the Feature Matrix to Gray Image (FM2GI) algorithm is
proposed to convert a single feature relationship of traffic accident’s data into gray images
containing combination relationships in parallel as the input variables for the model.
Moreover, experiments demonstrated that the proposed model for traffic accident’s severity
prediction had better performance. Top features affecting accidental severity include distance,
temperature, wind_Chill, humidity, visibility, and wind direction. This study presents an
ensemble of machine learning and deep learning models by combining Random Forest and
Convolutional Neural Network called RFCNN for the prediction of road accident severity.
The performance of the model is compared with several base learner classifiers.

2. A DATA MINING FRAMEWORK TO ANALYZE ROAD ACCIDENT DATA

One of the key objectives in accident data analysis to identify the main factors associated
with a road and traffic accident. However, the heterogeneous nature of road accident data
makes the analysis task difficult. Data segmentation has been used widely to overcome this
heterogeneity of the accident data. In this paper, we proposed a framework that used K-modes

10
clustering technique as a preliminary task for segmentation of 11,574 road accidents on the
road network of Dehradun (India) between 2009 and 2014 (both included). Next, association
rule mining is used to identify the various circumstances that are associated with the
occurrence of an accident for both the entire data set (EDS) and the clusters identified by
K-modes clustering algorithm. The findings of cluster based analysis and entire data set
analysis are then compared. The results reveal that the combination of k mode clustering and
association rule mining is very inspiring as it produces important information that would
remain hidden if no segmentation has been performed prior to generating association rules.
Further a trend analysis has also been performed for each cluster and EDS accidents which
finds different trends in different clusters whereas a positive trend is shown by EDS. Trend
analysis also shows that prior segmentation of accident data is very important before
analysis.. Association rule mining is used to identify the various circumstances that are
associated with the occurrence of an accident for both the entire data set (EDS) and the
clusters identified by K-modes clustering algorithm. The findings of cluster based analysis
and entire data set analysis are then compared. The results reveal that the combination of k
mode clustering and association rule mining is very inspiring as it produces important
information that would remain hidden if no segmentation has been performed prior to
generating association rules.

3. A COMPARATIVE STUDY OF MACHINE LEARNING ALGORITHMS TO

PREDICT ROAD ACCIDENT SEVERITY

In this work the author developed a machine learning model for identifying the severity
of the accident by considering the human factors such as alcohol, drug, age, and gender are
often ignored when determining accident severity. In this work single and ensemble mode
machine learning (ML) methods compared their performance in terms of prediction accuracy,
precision, recall, F1 score, area under the receiver operator characteristic (AUROC). Road
accidents are a global issue that cause deaths and injuries besides other direct and indirect
losses. Countries and international organizations have designed technologies, systems, and
policies to prevent accidents. The use of big traffic data and artificial intelligence may help

11
develop a promising solution to predict or reduce the risk of road accidents. Most existing
studies examine the impact of road geometry, environment, and weather parameters on road
accidents. However, human factors such as alcohol, drug, age, and gender are often ignored
when determining accident severity. In this work, we considered various contributing factors
and their impact on the prediction of the severity of accidents. For this, we studied a set of
single and ensemble mode machine learning (ML) methods and compared their performance
in terms of prediction accuracy, precision, recall, F1 score, area under the receiver operator
characteristic (AUROC).

4. ACCIDENT DATA ANALYSIS TO DEVELOP TARGET GROUPS FOR

COUNTERMEASURES

This work analyzes the accident data to develop target groups for developing the
accident countermeasures. The High risk group, high accident severity groups and accident
involved clusters are identified and used to develop the accident countermeasures. The author
identified each category such as vehicle, driver, passenger, environmental factors and the
leading accident factors among them. All the high risk factors are clustered to create a
countermeasure for each category.

5. ROAD TRAFFIC ACCIDENTS INJURY DATA ANALYTICS

Road safety researchers working on road accident data have witnessed success in road
traffic accidents analysis through the application of data analytic techniques, though little
progress was made into the prediction of road injury. This paper applies advanced data
analytics methods to predict injury severity levels and evaluates their performance. The study
uses predictive modeling techniques to identify risk and key factors that contribute to accident
severity. The study uses publicly available data from the UK department of transport that
covers the period from 2005 to 2019. The paper presents an approach which is general
enough so that it can be applied to different data sets from other countries. The results
identified that tree based techniques such as XGBoost outperform regression based ones, such

12
as ANN. In addition to the paper, it identifies interesting relationships and acknowledges
issues related to quality of data.

6. A REVIEW OF DATA ANALYTIC APPLICATIONS IN ROAD TRAFFIC

SAFETY. PART 1: DESCRIPTIVE AND PREDICTIVE MODELING

This work presents a comprehensive review on data analytic methods in road safety.
Analytics models can be grouped into two categories: predictive or explanatory models that
attempt to understand and quantify crash risk and (b) optimization techniques that focus on
minimizing crash risk through route/path selection and rest-break scheduling. Their work
presented publicly available data sources and descriptive analytic techniques (data
summarization, visualization, and dimension reduction) that can be used to achieve
safer-routing and provide code to facilitate data collection/exploration by
practitioners/researchers. reduce the start-up burden of data collection and descriptive
analytics for statistical modeling and route optimization of risk associated with motor
vehicles. From a data-driven bibliometric analysis, we show that the literature is divided into
two disparate research streams: (a) predictive or explanatory models that attempt to
understand and quantify crash risk based on different driving conditions, and (b) optimization
techniques that focus on minimizing crash risk through route/path-selection and rest-break
scheduling.

Translation of research outcomes between these two streams is limited. To overcome this
issue, we present publicly available high-quality data sources (different study designs,
outcome variables, and predictor variables) and descriptive analytic techniques (data
summarization, visualization, and dimension reduction) that can be used to achieve
safer-routing and provide code to facilitate data collection/exploration by
practitioners/researchers. Then, the author done the statistical and machine learning models
used for crash risk modeling.

13
7. TRAFFIC ACCIDENT PREDICTION MODEL USING SUPPORT VECTOR
MACHINES WITH GAUSSIAN KERNEL

Road traffic accident prediction models play a critical role in the improvement of traffic
safety planning. The focus of this study is to extract key factors from the collected data sets
which are responsible for the majority of accidents. In this paper urban traffic accident
analysis has been done using support vector machines (SVM) with Gaussian kernel.
Multilayer perceptron (MLP) and SVM models were trained, tested, and compared using
collected data. The results of the study reveal that the proposed model has significantly higher
prediction accuracy as compared with traditional MLP approach. There is a good relationship
between the simulated and the experimental values. Simulations were carried out using
LIBSVM (library for support vector machines) integrated with octave. In this work, applied
support vector machines with different Gaussian kernel functions for crash to extract
important features related to accident occurrence. The paper compared neural networks with
support vector machines. The paper reported that SVMs are superior in accuracy. However,
the SVMs method has the same disadvantages of ANN in traffic accident severity prediction.

8. PREDICTING FREEWAY TRAFFIC CRASH SEVERITY USING

XGBOOST-BAYESIAN NETWORK MODEL WITH CONSIDERATION OF
FEATURES INTERACTION

This project proposes the XGBoost based framework which analyzes the relationship
between collision, time and environmental and spatial factors and fatality rate. Results show
that the proposed method has the best modeling performance compared with other machine
learning algorithms. The paper identified eight factors that have an impact on traffic fatality.
In the field of freeway traffic safety research, there is an increasing focus in studies on how to
reduce the frequency and severity of traffic crashes. Although many studies divide factors
into “human-vehicle-road-environment” and other dimensions to construct models which

14
show the characteristic patterns of each factor's influence on crash severity, there is still a lack
of research on the interaction effect of road and environment characteristics on the severity of
a freeway traffic crash. This research aims to explore the influence of road and environmental
factors on the severity of a freeway traffic crash and establish a prediction model towards
freeway traffic crash severity.

Firstly, the obtained historical traffic crash data variables were screened, and 11
influencing factors were summarized from the perspective of road and environment, and the
related variables were discretized. Furthermore, the XGBoost (eXtreme Gradient Boosting)
model was established, and the SHAP (SHapley Additive exPlanation) value was introduced
to interpret the XGBoost model; the importance ranking of the influence degree of each
feature towards the target variables and the visualization of the global influence of each
feature towards the target variables were both obtained.

Then, the Bayesian network-based freeway traffic crash severity prediction model was
constructed via the selected variables and their values, and the learning and prediction
accuracy of the model were verified. Finally, based on the data of the case study, the
prediction model was applied to predict the crash severity considering the interaction effect of
various factors in road and environment dimensions.

The results show that the characteristic variables of road side protection facility type
(RSP), road section type (LAN), central isolation facility (CIF), lighting condition (LIG), and
crash occurrence time (TIM) have significant effects on the traffic crash prediction model; the
prediction performance of the model considering the interaction of road and environment is
better than that of the model considering the influence of single condition; the prediction
accuracy of XGBoost-Bayesian Network Model proposed in this research can reach 89.05%.
The identification and prediction of traffic crash risk is a prerequisite for safety improvement,
and the model proposed and results obtained in this research can provide a theoretical basis
for related departments in freeway safety management.

15
9. PREDICTING TRAFFIC ACCIDENT SEVERITY USING MACHINE
LEARNING TECHNIQUES

Road accidents, harming countries' economies, national assets as well as people's lives,
are one of the major problems for countries. Thus, investigating contributing factors to the
accidents and developing an accurate accident severity prediction model is critical. Using the
traffic accident data collected in Austin, Dallas, and San Antonio city of Texas between 2011
and 2021, the primary contributing factors in crashes are probed and the performance of a
deep learning model and five different machine learning techniques, such as Logistic
Regression, XGBoost, Random Forest, KNN, and SVM, are investigated. The finding shows
that the Logistic Regression algorithm shows the best performance among the others with an
accuracy of 88% in classifying accident severity.

10. A STUDY ON ROAD ACCIDENT PREDICTION AND CONTRIBUTING

FACTORS USING EXPLAINABLE MACHINE LEARNING MODELS: ANALYSIS
AND PERFORMANCE

Road accidents are increasing worldwide and are causing millions of deaths each year.
They impose significant financial and economic expenses on society. Existing research has
mostly studied road accident prediction as a classification problem, which aims to predict
whether a traffic accident may happen in the future or not without exploring the underlying
relationships between the complicated factors contributing to road accidents.

A number of researches have been done to date to explore the importance of road
accident contributing factors in relation to road accidents and their severity, however, only a
few of those research have explored a subset of ensemble ML models and the New Zealand
(NZ) road accident dataset. Therefore, in this paper, they have evaluated a set of machine
learning (ML) models to predict road accident severity based on the most recent NZ road
accident dataset. We have also analyzed the predicted results and applied an explainable ML
(XML) technique to evaluate the importance of road accident contributing factors.

To predict road accidents with different injury severity, this work has considered
different ensembles of ML models, like Random Forest (RF), Decision Jungle (DJ), Adaptive

16
Boosting (AdaBoost), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting
Machine (L-GBM), and Categorical Boosting (CatBoost). New Zealand road accident data
from 2016 through 2020 obtained from the New Zealand Ministry of Transport is used to
perform this study. The comparison results show that RF is the best classifier with 81.45%
accuracy, 81.68% precision, 81.42% recall, and 81.04% of F1-Score.

Next, they have employed the Shapley value analysis as an XML technique to interpret
the RF model performance at global and local levels. While the global level explanation
provides the rank of the features’ contribution to severity classification, the local one is for
exploring the use of features in the model. Furthermore, the Shapley Additive exPlanation
(SHAP) dependence plot is used to investigate the relationship and interaction of the features
towards the target variable prediction.

Based on the findings, it can be said that the road category and number of vehicles
involved in an accident significantly impact injury severity. The identified high-ranked
features through SHAP analysis are used to retrain the ML models and measure their
performance. The result shows 6%, 5%, and 8%, increase, respectively, in the performances
of DJ, AdaBoost, and CatBoost models.

17
 CHAPTER 3

ROAD ACCIDENT SEVERITY ANALYSIS

3.1 METHODOLOGY

Fig 3.1 : Block diagram of the proposed system

This project classifies the road accident as slight or serious. The flow of the project is
shown in figure 3.1. This project consists of five main phases.

1. Conducting descriptive study on the accident data

2. Pre-processing the data using grouping and label encoding

3. Building the machine learning classification model

4. Performance evaluation

5. Developing intelligent web portal for accident severity classification 6

18
 Fig 3.2 : Block diagram of web portal

The steps involved in the intelligence web portal are described in the figure 3.2.

3.1.1 Dataset

This project was implemented using the Accident Severity dataset from the kaggle. The
dataset contains 3057 accident samples. Each sample contains 14 predictive attributes and 1
target attribute. The dataset contains the accident data from 01-01-2009 to 31- 12-2009. There
are no null values present in the dataset. The target attribute contains two values: slight
accident and serious accident. The dataset contains 321 serious data and 2736 slight accident
data.

The figure 3.3 describes the predictive attributes that are available in the accident data and
its description.

19
 Table 3.1 Accident Dataset Attributes

3.1.2 Data Preprocessing

Data preprocessing is a data mining technique which is used to transform the raw data in a
useful and efficient format. Data preprocessing transforms the data into a format that is more
easily and effectively processed in data mining, machine learning and other data science
tasks. The techniques are generally used at the earliest stages of the machine learning and AI
development pipeline to ensure accurate results.

Three data preprocessing tasks were conducted in the accident data

1. Data reduction

2. Encoding the categorical data

3. Dropping the unwanted columns

Some attribute values can be combined into a single variable. By grouping those values into
the categories will help to reduce the number of attributes. The same thing is applicable to

20
reduce the number of categories in an attribute. Consider the ‘Type of vehicle’ column in the
accident data [ M/cycle 50cc and under', 'Motorcycle over 125cc and up to 500cc',
'Motorcycle over 50cc and up to 125cc', 'Motorcycle over 500cc ] these are all the different
kinds of two wheeler motorcycles. Hence, all the data can be grouped into a single category
as ‘Motorcycle’.

ii. Data Encoding

Encoding categorical data is a process of converting categorical data into integer format
so that the data with converted categorical values can be provided to the different models. In
the field of data science, before going for modeling, data preparation is a mandatory task.
Label Encoding is a popular encoding technique for handling categorical variables. In this
technique, each label is assigned a unique integer based on alphabetical ordering. The figure
3.3 describe how the label encoding performed on the accident data.

Table 3.2 Label Encoding

iii. Drop Unwanted Columns
Some of the attributes are usually stored for reference purposes. Those kinds of attributes are
not used for the prediction. That data might reduce the accuracy of the classifier. Hence,
those columns are removed from the attribute list.

21
3.1.3 Build machine learning model

Four machine learning classifiers such as Decision tree learning, K-Nearest neighbor,
Random forest classifier and the Gradient boosting classifiers are used to develop machine
learning models. These models are imported using the Sci-kit learn machine learning library.

3.1.4 Training and Testing

The pre-processed data splitted into two sets. 70% of the data considered as the training
data and the rest 30% used to test the model.

3.1.5 Result Analysis

The machine learning models are evaluated using the standard metrics such as accuracy,
precision, recall and ROC curve. Based on the metrics of each classifier, the comparative
study was conducted over the models. The result shows that the ensembler can give the best
accuracy among the traditional single machine learning classifiers.

3.2 GRAPHICAL USER INTERFACE

To deploy the model in the web browser the user interface was designed using HTML,
CSS and Javascript. Flask framework used to provide the interface between the Mysql
database and the web browser.

3.2.1 Save the machine learning model

To save the model, we simply need to pass the model object to Pickle’s dump() function.
This will serialize the object and convert it into a ”byte stream” that can be stored in the
model.pkl file. From the result analysis the Gradient Boosting Classifier provides better
accuracy.

22
3. 3 TOOLS USED

The tools used for the project:

● ∙Python
● Jupyter Notebook
● Flask
● HTML
● CSS
● JavaScript
● MySql
● Spyder

3. 3. 1 Python

Python is a high-level, interpreted, general-purpose programming language. Its design

philosophy emphasizes code readability with the use of significant indentation. Python is
dynamically-typed and garbage-collected. It supports multiple programming paradigms,
including structured (particularly procedural), object-oriented and functional programming. It
is often described as a "batteries included" language due to its comprehensive standard.

Guido van Rossum began working on Python in the late 1980s as a successor to the ABC
programming language and first released it in 1991 as Python 0.9.0.[35] Python 2.0 was
released in 2000 and introduced new features such as list comprehensions, cycle-detecting
garbage collection, reference counting, and Unicode support. Python 3.0, released in 2008,
was a major revision that is not completely backward-compatible with earlier versions.

23
i. Pandas

Pandas is a software library written for the Python programming language for data
manipulation and analysis. In particular, it offers data structures and operations for
manipulating numerical tables and time series. It is free software released under the
three-clause BSD license.

ii. Scikit Learn

Scikit-learn is a free machine learning library for Python. It features various algorithms
like support vector machine, random forests, and k-neighbors, and it also supports Python
numerical and scientific libraries like NumPy and SciPy.

iii. Collections

Collections in Python are containers that are used to store collections of data, for
example, list, dict, set, tuple etc. These are built-in collections. The collections module
provides alternatives to built-in container data types such as list, tuple and dict. Several
modules have been developed that provide additional data structures to store collections of
data

iv. MatplotLib

It is an amazing visualization library in Python for 2D plots of arrays. It is a

multiplatform data visualization library built on NumPy arrays and designed to work with the
broader SciPy stack. One of the greatest benefits of visualization is that it allows us visual
access to huge amounts of data in easily digestible visuals. Matplotlib consists of several
plots like line, bar, scatter, histogram etc.

24
v. NumPy

NumPy is a general-purpose array-processing package. It provides a high performance

multidimensional array object, and tools for working with these arrays. It is the fundamental
package for scientific computing with Python.

3.3.2 Jupyter Notebook

The Jupyter Notebook is an open-source web application for creating and sharing
documents with live code, equations, visualizations, and text. Jupyter Notebook is
administered by the Project Jupyter team. The Jupyter Notebooks project is a spin-off of the
IPython project, which previously had its own IPython Notebook project. Jupyter derives its
name from the three primary programming languages it supports: Julia, Python, and R.
Jupyter ships with the IPython kernel, which enables you to write your programmes in
Python, but you can also use more than 100 other kernels. Jupyter notebooks are intended to
provide a more user-friendly interface for code used in digitally-supported research or
education.

Jupyter, an open-source environment compatible with a variety of programming

languages, has gained traction in a variety of fields.It is useful for documenting code,
teaching programming languages, and providing students with a space to easily experiment
with provided examples. Jupyter Notebooks can be executed in two major environments:
Jupyter Notebook and the more recent JupyterLab. Jupyter Notebook is widely used and
well-documented; it provides a simple file browser and an environment for creating, editing,
and executing notebooks. Jupyter Lab is more complex and resembles an Integrated
Development Environment in its user interface.

The Jupyter Notebook is not only useful for teaching and learning programming
languages like Python, but also for sharing data. Google and Microsoft each offer their own

25
version of the Notebook, which can be used to create and share documents at Google
Colaboratory and Microsoft Azure Notebooks, respectively. JupyterLab incorporates the
Jupyter Notebook into a browser-based Integrated Development type Editor. JupyterLab can
be viewed as an advanced version of Jupyter Notebook. In addition to Note, JupyterLab
allows you to run terminals, text editors, and code consoles in your browser.

3.3.3 Flask

Flask is an API of Python that allows us to build up web-applications. It was developed

by Armin Ronacher. Flask’s framework is more explicit than Django’s framework and is also
easier to learn because it has less base code to implement a simple web Application. A
Web-Application Framework or Web Framework is the collection of modules and libraries
that helps the developer to write applications without writing the low-level codes such as
protocols, thread management, etc. Flask is based on the WSGI(Web Server Gateway
Interface) toolkit and Jinja2 template engine.

i. HTTP Methods

GET - This is used to send the data in and without encryption of the form to the server.

POST - Sends the form data to the server. Data received by POST method is not cached by
the server.

ii. Routing

Nowadays, the web frameworks provide routing techniques so that users can remember
the URLs. It is useful to access the web page directly without navigating from the Home
page. It is done through the following route() decorator, to bind the URL to a function.

26
iii. Handling Static Files

A web application often requires a static file such as javascript or a CSS file to render the
display of the web page in the browser. Usually, the web server is configured to set them, but
during development, these files are served as static folders in your package or next to the
module.

3.3.4 HTML

HTML is the standard markup language for documents intended for display in a web
browser. It can be aided by Cascading Style Sheets (CSS) and scripting languages such as
JavaScript. Web browsers receive HTML files from a web server or local storage and convert
them into multimedia web pages. HTML describes the semantic structure of a web page and
originally included hints for the document’s appearance.

HTML elements are the fundamental constituents of HTML pages. Images and other
objects, such as interactive forms, can be embedded in the rendered page using HTML
constructs. HTML enables the creation of structured documents by assigning structural
semantics to text elements such as headings, paragraphs, lists, links, and other elements.

3.3.5 CSS

CSS is a style sheet language that is used to describe the presentation of a document
written in a markup language such as HTML. In addition to HTML and JavaScript, CSS is a
fundamental technology for the World Wide Web. CSS is designed to separate presentation
from content, including layout, colors, and fonts. This separation can improve content
accessibility; provide greater flexibility and control in the specification of presentation

27
characteristics; enable multiple web pages to share formatting by specifying the relevant CSS
in a separate.css file, which reduces complexity and repetition in the structural content; and
enable the.css file to be cached to improve page load speed for pages that share the file and
its formatting.

3.3.6 JavaScript

JavaScript, also known as JS, is a programming language that, along with HTML and
CSS, is one of the core technologies of the World Wide Web. Over 97 percent of websites use
JavaScript for client-side web page behavior, with third-party libraries frequently
incorporated. All major web browsers include a dedicated JavaScript engine for executing
code on user devices. JavaScript is a high-level, often just-in-time, ECMAScript-compliant,
compiled programming language. It features dynamic typing, prototype-based object
orientation, and functions with first-class status. Event-driven, functional, and imperative
programming styles are supported. APIs for working with text, dates, regular expressions,
standard data structures, and the Document Object Model are available (DOM). Initially,
JavaScript engines were only used in web browsers, but they are now integral components of
many servers and applications.

3.3.7 MySql

MySQL is an open-source relational database management system (RDBMS). Its name

is a combination of "My", the name of co-founder Michael Widenius's daughter and "SQL",
the abbreviation for Structured Query Language. A relational database organizes data into
one or more data tables in which data may be related to each other; these relations help
structure the data.

SQL is a language programmers use to create, modify and extract data from the
relational database, as well as control user access to the database. In addition to relational
databases and SQL, an RDBMS like MySQL works with an operating system to implement a

28
relational database in a computer's storage system, manages users, allows for network access
and facilitates testing database integrity and creation of backups.

3.3.8 Spyder

Spyder is an open-source cross-platform integrated development environment (IDE) for

scientific programming in the Python language. Spyder integrates with a number of
prominent packages in the scientific Python stack, including NumPy, SciPy, Matplotlib,
pandas, IPython, SymPy and Cython, as well as other open-source software.

3.3.9 Weather Stack API

API covers global weather data across the board — from a multi-year history all the way
to live information and accurate weather forecasts. The first step to using the API is to
authenticate with your weatherstack account's unique API access key, which can be found in
your account dashboard after registration. To authenticate with the API, simply use the base
URL below and pass your API access key to the API's access_key parameter.

3.3.10 Flask-Session
Flask-Session is an extension for Flask that supports Server-side Session to the
application. The Session is the time between the client logs in to the server and logs out of the
server. The data that is required to be saved in the Session is stored in a temporary directory
on the server. The data in the Session is stored on the top of cookies and signed by the server
cryptographically. Each client will have their own session where their own data will be stored
in their session.

3. 4 MACHINE LEARNING ALGORITHM

Machine learning is a subset of AI, which enables the machine to automatically learn
from data, improve performance from past experiences, and make predictions. Machine

29
learning contains a set of algorithms that work on a huge amount of data. Data is fed to these
algorithms to train them, and on the basis of training, they build the model & perform a
specific task.

These ML algorithms help to solve different business problems like Regression,

Classification, Forecasting, Clustering, and Associations, etc. Based on the methods and way
of learning, machine learning is divided into mainly four types, which are:

1. Supervised Machine Learning

2. Unsupervised Machine Learning
3. Semi-Supervised Machine Learning
4. Reinforcement Learning

I. Supervised Machine Learning

As its name suggests, Supervised machine learning is based on supervision. It means in

the supervised learning technique, we train the machines using the "labeled" dataset, and
based on the training, the machine predicts the output. Here, the labeled data specifies that
some of the inputs are already mapped to the output. More precisely, we can say; first, we
train the machine with the input and corresponding output, and then we ask the machine to
predict the output using the test dataset.

Let's understand supervised learning with an example. Suppose we have an input dataset
of cats and dog images. So, first, we will provide the training to the machine to understand the
images, such as the shape & size of the tail of cat and dog, Shape of eyes, color, height (dogs
are taller, cats are smaller), etc. After completion of training, we input the picture of a cat and
ask the machine to identify the object and predict the output. Now, the machine is well
trained, so it will check all the features of the object, such as height, shape, color, eyes, ears,
tail, etc., and find that it's a cat. So, it will put it in the Cat category. This is the process of how

30
 the machine identifies the objects in Supervised Learning. The main goal of the supervised
learning technique is to map the input variable(x) with the output variable(y). Some real-world
applications of supervised learning are Risk Assessment, Fraud Detection, Spam filtering, etc.

II. Unsupervised Machine Learning

Unsupervised learning is different from the Supervised learning technique; as its name
suggests, there is no need for supervision. It means, in unsupervised machine learning, the
machine is trained using the unlabeled dataset, and the machine predicts the output without
any supervision.

In unsupervised learning, the models are trained with the data that is neither classified nor
labeled, and the model acts on that data without any supervision. The main aim of the
unsupervised learning algorithm is to group or categorize the unsorted dataset according to the
similarities, patterns, and differences. Machines are instructed to find the hidden patterns from
the input dataset.

Let's take an example to understand it more precisely; suppose there is a basket of fruit
images, and we input it into the machine learning model. The images are totally unknown to
the model, and the task of the machine is to find the patterns and categories of the objects. So,
now the machine will discover its patterns and differences, such as color difference, shape
difference, and predict the output when it is tested with the test dataset.

III. Semi Supervised Machine Learning

Semi-Supervised learning is a type of Machine Learning algorithm that lies between

Supervised and Unsupervised machine learning. It represents the intermediate ground between
Supervised (With Labeled training data) and Unsupervised learning (with no labeled training
data) algorithms and uses the combination of labeled and unlabeled data sets during the
training period.

31
 Although Semi-supervised learning is the middle ground between supervised and
unsupervised learning and operates on the data that consists of a few labels, it mostly consists
of unlabeled data. As labels are costly, but for corporate purposes, they may have few labels. It
is completely different from supervised and unsupervised learning as they are based on the
presence & absence of labels.

To overcome the drawbacks of supervised learning and unsupervised learning algorithms,

the concept of Semi-supervised learning is introduced. The main aim of semi-supervised
learning is to effectively use all the available data, rather than only labeled data like in
supervised learning. Initially, similar data is clustered along with an unsupervised learning
algorithm, and further, it helps to label the unlabeled data into labeled data. It is because
labeled data is a comparatively more expensive acquisition than unlabeled data.

IV. Reinforcement Learning

Reinforcement learning works on a feedback-based process, in which an AI agent (A

software component) automatically explores its surroundings by hitting & trail, taking action,
learning from experiences, and improving its performance. Agents get rewarded for each
good action and get punished for each bad action; hence the goal of reinforcement learning
agents is to maximize the rewards.

3.4.1 Decision Tree Learning

The Decision Tree algorithm belongs to the family of supervised learning algorithms.
Unlike other supervised learning algorithms, the decision tree algorithm can be used for
solving regression and classification problems too. The goal of using a Decision Tree is to
create a training model that can be used to predict the class or value of the target variable by
learning simple decision rules inferred from prior data(training data).

32
In Decision Trees, for predicting a class label for a record we start from the root of the tree.
We compare the values of the root attribute with the record’s attribute. On the basis of
comparison, we follow the branch corresponding to that value and jump to the next node.

Terminologies Of The Decision Trees

Root Node: It represents the entire population or sample and this further gets divided into
two or more homogeneous sets.

Splitting: It is a process of dividing a node into two or more sub-nodes.

Decision Node: When a sub-node splits into further sub-nodes, then it is called the decision
node.

Leaf / Terminal Node: Nodes that do not split are called Leaf or Terminal nodes.

Pruning: When we remove sub-nodes of a decision node, this process is called pruning. You
can say the opposite process of splitting.

Branch / Sub-Tree: A subsection of the entire tree is called branch or sub-tree.

Parent and Child Node: A node, which is divided into sub-nodes is called a parent node of
sub-nodes whereas sub-nodes are the child of a parent node.

The figure 3.5 describes the general structure of the decision tree.

33
 Fig 3.5 : Decision Tree
Steps

● It begins with the original set S as the root node.

● On each iteration of the algorithm, it iterates through the very unused attribute of the
set S and calculates Entropy(H) and Information gain(IG) of this attribute. ∙ It then selects the
attribute which has the smallest Entropy or Largest Information gain.
● The set S is then split by the selected attribute to produce a subset of the data. ∙ The
algorithm continues to recur on each subset, considering only attributes never selected
before.

Entropy

Entropy is a measure of the randomness in the information being processed. The higher
the entropy, the harder it is to draw any conclusions from that information. Flipping a coin is
an example of an action that provides information that is random.

…(1)

Where, Pi = Probability of randomly selecting an example in class I

34
Information Gain

Information gain or IG is a statistical property that measures how well a given attribute
separates the training examples according to their target classification. Constructing a
decision tree is all about finding an attribute that returns the highest information gain and the
smallest entropy.

Information Gain = 1 – Entropy …(2)

3.4.2 K-Nearest Neighbor

K-Nearest Neighbor is one of the simplest Machine Learning algorithms that utilizes the
Supervised Learning technique.It assumes the similarity between the new case/data and
existing cases and places the new case in the category that is the most similar to the existing
categories.The K-NN algorithm stores all available data and classifies a new data point on
the basis of similarity. This implies that when new data becomes available, the K- NN
algorithm can easily classify it into a suitable category.It can be used for both Regression and
Classification, although Classification is its primary application.

Fig 3.6 : K-nearest Neighbor

35
 K-NN is a non-parametric algorithm, meaning it makes no assumptions about the data it
is analysing.It is also referred to as a lazy learner algorithm because it does not immediately
learn from the training set. Rather, it stores the dataset and, at the time of classification,
performs an action on the dataset. During the training phase, the system simply stores the
dataset, and when it receives new data, it classifies it into a category that is highly similar to
the original category.

3.4.3 Random Forest Classifier

Popular machine learning algorithm that belongs to the supervised learning technique is
Random Forest. It is applicable to both Classification and Regression problems in Machine
Learning. It is based on ensemble learning, which is the process of combining multiple
classifiers to solve a complex problem and improve the model’s performance.

Fig 3.7 : Random Forest Classifier

Since the random forest combines multiple trees to predict the class of the dataset, it is
possible that some decision trees may predict the correct output, while others may not. But
together, all the trees predict the correct output. Therefore, below are two assumptions for a
better Random forest classifier: There should be some actual values in the feature variable of

36
the dataset so that the classifier can predict accurate results rather than a guessed result. The
predictions from each tree must have very low correlations.

Steps :

Random Forest works in two-phase first is to create the random forest by combining N
decision tree, and second is to make predictions for each tree created in the first phase.
Step 1: Select random K data points from the training set.
Step 2: Build the decision trees associated with the selected data points (Subsets). Step 3:
Choose the number N for decision trees that you want to build.
Step 4: Repeat Step 1 & 2.
Step 5: For new data points, find the predictions of each decision tree, and assign the new
data points to the category that wins the majority votes.

3.4.4 Gradient Boosting Classifier

Gradient boosting algorithm is one of the most powerful algorithms in the field of
machine learning. As we know that the errors in machine learning algorithms are broadly
classified into two categories i.e. Bias Error and Variance Error. As gradient boosting is one
of the boosting algorithms it is used to minimize bias error of the model.

Fig 3.8 : Gradient Boosting Classifier

37
 Gradient boosting algorithms can be used for predicting not only continuous target
variables (as a Regressor) but also categorical target variables (as a Classifier). When it is
used as a regressor, the cost function is Mean Square Error (MSE) and when it is used as a
classifier then the cost function is Log loss.

Steps:
Step 1: Calculate the average of the target label
Step 2: Calculate the residuals : residual = actual value – predicted value
Step 3: Construct a decision tree
Step 4: Predict the target label using all of the trees within the ensemble Step 5: Compute the
new residuals
Step 6: Repeat steps 3 to 5 until the number of iterations matches the number specified by
the hyperparameter (i.e. number of estimators)
Step 7: Once trained, use all of the trees in the ensemble to make a final prediction as to the
value of the target variable

38
 CHAPTER 4

RESULTS ANALYSIS

4.1 DATA PREPROCESSING RESULT

Figure 4.1 describes the dataset structure before preprocessing. The dataset downloaded
from the kaggle and data preprocessing conducted inorder to transform into the machine
readable form.

Fig 4.1 : Dataset before preprocessing

Figure 4.2 describes the dataset after performing the pre-processing such as data
reduction, label encoding and dropping unwanted attributes. The dataset will be transformed
as shown in figure. The data is stored as a csv file and used in the training and testing phases.
30% data used for the testing purpose and the 70% of the data utilized in the training purpose.
Each model has been trained and tested using the same data.

The label encoding outperforms while converting the text data into the numerical form.
Since some attributes contain multiple values under the same group of attributes and those are
grouped based on its similarity.

39
 Fig 4.2 : Dataset after preprocessing
4.2 PERFORMANCE METRICS

The following metrics are used to evaluate the performance of the classification.

4.2.1 Confusion Matrix

A confusion matrix, also known as an error matrix, is a summarized table used to assess
the performance of a classification model. The number of correct and incorrect predictions
are summarized with count values and broken down by each class. The confusion matrix is as
shown in figure 4.3.

Table 4.1 Confusion matrix

Positive (P): Observation is positive

Negative (N): Observation is not positive.

40
True Positive (TP): Outcome where the model correctly predicts the positive class. True
Negative (TN): Outcome where the model correctly predicts the negative class.

False Positive (FP): Also called a type 1 error, an outcome where the model incorrectly
predicts the positive class when it is actually negative.

False Negative (FN): Also called a type 2 error, an outcome where the model incorrectly
predicts the negative class when it is actually positive.

4.2.2 Accuracy

Accuracy is defined as the ratio of correctly predicted examples by the total examples.
Accuracy of the classification is calculated using Eq. (3)

Accuracy = TP + TN / (TP + TN + FP + FN) …(3)

4.2.3 Error Rate

Error rate is calculated by eliminating the accuracy from the total accuracy

Error rate = 1 – Accuracy …(4)

4.2.4 Precision

Precision is also called Positive predictive value. Precision is also known as positive
predictive value and is the proportion of relevant instances among the retrieved instances.The
ratio of correct positive predictions to the total predicted positives. Precision of the
classification is calculated using Eq. (5)

Precision = TP / (TP + FP) …(5)

41
4.2.5 Recall

Recall also called Sensitivity, Probability of Detection, True Positive Rate. The ratio of
correct positive predictions to the total positive examples. Recall of the classification is
calculated using Eq. (6)

Recall = TP / (TP + FN) …(6)

4.2.6 Specificity
Specificity is defined as the proportion of actual negatives, which got predicted as the
negative. Specificity of the classification is calculated using Eq. (7)

Specificity = TN / (TN + FP) …(7)

4.2.7 ROC Curve

A ROC curve (receiver operating characteristic curve) graph shows the performance of a
classification model at all classification thresholds. It plots two Parameters namely True
Positive (TPR) rate and False positive rate (FPR).
TPR(Eq. 8) and FPR(Eq. 9) of the classification is calculated as follows
True Positive Rate = TP/ ( FP + TN ) ... (8)
False positive Rate = FP / ( FP + TN ) …(9)

4.2.8 AUC

AUC stands for Area under the ROC Curve. A perfect classifier would have an AUC of 1.
Usually, if your model behaves well, you obtain a good classifier by selecting the value of the
threshold that gives TPR close to 1 while keeping FPR near 0.

42
4.2.9 Performance comparison over the classifiers

The figure 4.4 describes the accuracies that are gained by the considered classifiers.
Gaining high accuracy is the important thing to be noticed. The Decision tree learning has got
89% of accuracy. The KNN got 89.4 % of accuracy and worked better than the KNN. Nearly
the Random forest algorithm got 90% of accuracy. The best accuracy gained by the Gradient
boosting model. Since it is an ensemble classifier, it employs many weak learners in order to
learn the severity classification. Hence, its performance is better than the classifications that
are considered.

Fig 4.3 Accuracy score

The figure 4.5 describes the recall that is achieved by the considered classification
algorithms. Recall score is used to measure the model performance in terms of measuring the
count of true positives in a correct manner out of all the actual positive values.
Precision-Recall score is a useful measure of success of prediction when the classes are very
imbalanced.

43
 Fig 4.4 Recall score

From the figure 4.4 the Gradient boosting classifier gives the best recall results than other
considered classifiers. This mimics the gradient boosting classifier and can give the best recall
predictions while considering the road accident data. This shows that the gradient boosting
classifiers can work well even if the data is imbalanced. Since it is the binary classification
the gradient boosting classifier outperforms while considering the accuracy and recall.

The ROC curve is another performance metric used to evaluate the machine learning
model. An ROC curve (receiver operating characteristic curve) is a graph showing the
performance of a classification model at all classification thresholds. This curve plots two
parameters: True Positive Rate. False Positive Rate.

44
 DT Classifier KNN Classifier

RF Classifier GB Classifier

Table 4.2 ROC Curve

The above table 4.2 describes the ROC Curves that are achieved by the considered classifier.
The gradient boosting classifier has its curve from less false positive rate.

45
4.3 GRAPHICAL USER INTERFACE

The main objective of developing a user interface is to deploy the extracted machine
learning model in a website. By this implementation any end users without any technical
knowledge about machine learning also can use the model to predict the accident severity.

The application consists of three phases.

● Front End
● Programming Interface
● Database Server
i. Front End
The front end of this project is developed using HTML, CSS and JavaScript. The friend
end designed with the below functionalities
1. User Registration
2. User Login
3. Accident Severity Detection in the User Live Location
4. Accident Severity Detection in the User Preferred Location
5. Accident Severity Detection between two Locations
6. Downloading the Prediction as Report
7. Send the Report as Mail
ii. Programming Interface
Flask is used as a backend . It is used to provide the interface between the web browser
and the MySql database. The user details, accident details and location details are saved and
retrieved from the database

46
iii. Database Server
Xampp server is used to provide a server based mechanism for the MySql Database. The
below are some of the images from the intelligent accident severity detection portal.

Fig 4.5 User Authentication

Fig 4.6 Live Accident severity Detection

47
 Fig 4.7 Live Accident severity Detection Result

The web application has been tested with various test data that are available and
unavailable in the training and test dataset. The system predicts accurately whatever the data
provided. Since it is a live data based prediction, it works well in various geographical
locations also.

48
 CHAPTER 5

CONCLUSION AND FUTURE SCOPE

5.1 CONCLUSION

Road accidents are one of the most regrettable hazards in this hectic world. Road
accidents lead to numerous casualties, injuries, and fatalities each year, as well as significant
economic losses. Predicting the accident severity is one of the major tasks. The Gradient
boosting model outperforms while considering the accident data and it achieved 93% of
accuracy. Number of Vehicles, Road Class, Road Surface, Lighting Conditions, Spot Weather
Conditions, Casualty Class, Sex of Casualty, Age of Casualty, Type of Vehicle are used to
predict the accident severity. This is extraordinarily beneficial for the highway authorities,
police departments and for journalists. The key applications of this project are Early accident
severity prediction, No expert knowledge required, Can access the model anytime and
anywhere, Can access the previous predictions, Can send mail immediately to the respective
authority.

5.2 FUTURE SCOPE

The methodology can be used for pre prediction also. Whenever the driver, traveler or
passenger starts a journey in a particular area they can predict the accidents happening in that
area and the severity of the accident. This can be further used to identify the risk factors,
countermeasures. The previous predicted data also can be used to predict the future accident
severities and to improve the efficiency of the model.

49
REFERENCES

- Mubariz mansoor, Muhammad umar , Saima sadiq , Abid isaq, Saleem ullah, Hamza, and
Carmen, "RFCNN: Traffic Accident Severity Prediction Based on Decision Level Fusion of
Machine and Deep Learning Model", Digital Object Identifier
10.1109/ACCESS.2021.3112546.

- Sachin Kumar and Durga Toshniwal, "A data mining framework to analyze road accident
data", DOI 10.1186/s40537-015-0035-y

- Shakil Ahmed, Md Akbar Hossain, Md Mafijul Islam Bhuiyan, Sayan Kumar Ray, "A
Comparative Study of Machine Learning Algorithms to Predict Road Accident Severity",
978-1-6654-6667-7/21/$31.00 ©2021 IEEE DOI 10.1109/IUCC-CIT-
DSCI-SmartCNS55181.2021.00069

- Accident Data Analysis to Develop Target Groups For Countermeasures, Max Cameron

- Mohamed K Nour, Atif Naseer, Basem Alkazemi, Muhammad, "Road Traffic Accidents
Injury Data Analytics", (IJACSA) International Journal of Advanced Computer Science and
Applications, Vol. 11, No. 12, 2020

- A. Mehdizadeh, M. Cai, Q. Hu, M. A. A. Yazdi, N. Mohabbati-Kalejahi, A. Vinel, S. E.

Rigdon, K. C. Davis, and F. M. Megahed, “A review of data analytic based applications in
road traffic safety. Part 1: Descriptive and predictive modeling,” Sensors (Switzerland), vol.
20, no. 4, pp. 1–24, 2020.

- Q. Hu, M. Cai, N. Mohabbati-Kalejahi, A. Mehdizadeh, M. A. A. Yazdi, A. Vinel, S. E.

Rigdon, K. C. Davis, and F. M. Megahed, “A review of data analytic applications in road
traffic safety. Part 2: Prescriptive modeling,” Sensors (Switzerland), vol. 20, no. 4, pp. 1–19,
2020.

50
- J. Ma, Y. Ding, J. C. Cheng, Y. Tan, V. J. Gan, and J. Zhang, “Analyzing the Leading
Causes of Traffic Fatalities Using XGBoost and Grid-Based Analysis: A City Management
Perspective,” IEEE Access, vol. 7, pp. 148 059–148 072, 2019.

- N. Zagorodnikh, A. Novikov, and A. Yastrebkov, “Algorithm and software for identifying

accident-prone road sections,” Transp. Res. Procedia, vol. 36, pp. 817– 825, 2018. [Online].
Available: https://doi.org/10.1016/j.trpro.2018.12.074.

- L. G. Cuenca, E. Puertas, N. Aliane, and J. F. Andres, “Traffic Accidents Classification and

Injury Severity Prediction,” in 2018 3rd IEEE Int. Conf. Intell. Transp. Eng. ICITE 2018,
2018, pp. 52–57

51