DEPARTMENT OF ELECTRONICS AND
TELECOMMUNICATION ENGG
MAJOR PROJECT
On
COVID – 19 Outbreak Prediction using
Machine Learning
Under the Guidance of
Dr. Avik Banerjee
7TH SEMESTER
MANISH KUMAR SINGH— 1901227508
SHREYASH SRIVASTAVA— 1901227527
SWAYANSHA MOHANTY—
1901227546
PROJECT SUMMARY
Key Achievements in 6th Semester
Successfully completed Automatic License Plate Detection Project Using Image Processing
Tools.
Any Challenges/ Issues faced
The preprocessing of data was a Major Challenge.
Work to be done in 7th Semester
To study the research paper related Covid_19 .
To demonstrate the Trends of Covid_19 in different states in India.
To use the best suitable Algorithm for predicting the future forecast .
2
Agenda / Contents
01 Introduction
02 Motivation
03 Literature Survey
04 Problem Statement
05 Methodology
06 Results
07 Conclusion
08 Future Work
09 ReferencesC. V. Raman Global University, Odisha
Motivation
The outbreak of COVID-19 is developing into a major international crisis, and it’s starting to
influence important aspects of daily life.
Finding out how to predict and estimate the pandemic trend precisely is of huge necessity
because COVID-19 has made the world economy in a recession and deprived over millions
thousand lives.
To control the spread of COVID-19, this unprecedented public health crisis, it is of a great
necessity to predict the trend of the epidemic as precise as possible, providing valuable
references to policymaking, and resource.[1] allocating.
Fig2: Represents the devastation impact
Fig1: Represents the new lifestyle after covid -19
of covid-19 on economy
1
C. V. Raman Global University, Odisha
Introduction
Corona Virus disease (COVID-19) is an infectious disease caused by a newly
discovered virus, which emerged in Wuhan, China in December of 2019.[1]
The COVID-19 has largely impacted on all the sectors like economy, education,
healthcare, logistics and mental health of people.
With the help of predictive analysis and random forest algorithm, we can predict
the future cases which might be helpful for taking much better preventive
measures and precautions.
Our objective is to build a Machine Learning model that predicts the spread of
the covid-19 virus in upcoming days.
1
C. V. Raman Global University, Odisha
Literature Review
SL. TITLE/ AUTHOR METHODOLOGY REMARKS
No CONFERENCE/
YEAR
ADVANTAGES OTHER REMARK
DISADVANTAGES
1. TITLE - Zhihao Yang, Kang’an There are two Methodology 1. Training on this 1. The results from The model can be
Chen used One is traditional network converges FCNN and LSTM do a better model
Machine Learning
regression models, through adding
Methods on COVID-19 at around 50 epochs not meet our
including Lasso Regression, new features,
Situation Prediction with a validating expectations due which can reflect
Ridge Regression, and accuracy of 67%. to data deficiency, government
Multivariate Polynomial Testing a set of real with the highest implementation,
YEAR - 2020 Regression. data with an input of accuracy of 37%, local culture, and
Another is the models (.329, .222, .449). exploring new environmental
based on neural networks, Numbers in approaches to factor. However,
CONFERENCE including FCNN parenthesis denote make the network this kind of new
Name – (FullyConnected Neural positive ratio, more robust is features may not
International Network), LSTM (Long negative ratio, and significant. Rather be quantified
Conference on Short-Term Memory), and a neutral ratio, than optimizing easily.
contemporay classification model built by respectively. network structure,
us. This classification model introducing more
computing bases on twitter data to train
sentimental analysis and network is more
uses the overall percentage feasible.
PUBLISHER – of different emotion types
to predict a plausible range
IEEE
of daily increase numbers.
[2]
C. V. Raman Global University, Odisha
Literature Review
SL. TITLE/CONFERENCE/ AUTHOR METHODOLOGY REMARKS
No YEAR
ADVANTAGE DISADVANTAGE OTHER REMARK
2. TITLE - Safa Bhari, The methodologies involved [Link] Deep [Link] the The paper
Deep Learning for covid-19 Moetez Kdayem, for model training and testing learning model can model accuracy focuses on to
predictions Nesrine Zoghlami Artificial Neural Networks successfully fit the gets failed due to deploy a
that are basically the dynamic of different factors as Recurrent Neural
computational mechanism evolution of if the new variant Network
CONFERENCE- able to acquire, represent observed COVID-19 of covid-19 comes architecture
International Conference on and compute from one recovered cases into picture it called Long-Short
Advance System and multivariate space of and provide involved different Term memory
emergent Technologies. information to another. accurate prediction other information (LSTM) Neural
Recurrent Neural Network is a Network for the
in the dataset.
type of Artificial Neural next covid
Network designed to process recovered cases
YEAR - 2021 and analyse streams of in USA.
sequential data.[3]
PUBLISHER –
IEEE
1
C. V. Raman Global University, Odisha
Literature Review
SL. TITLE/JOURNAL/YEAR AUTHOR METHODOLOGY
No REMARKS
ADVANTAGE OTHER REMARK
DISADVANTAGE
3. TITLE – Ashish U 1. In the proposed model there 1. The paper done 1. The paper The evaluation
Prediction of Covid-19 Mandayam, are two stages: training and complete majorly focuses of the models
pandemic based on Rakshith.A.C, evaluation. Before training, analysis on the on the performance is
regression. Siddesha S, S K dataset is pre-processed by prediction of predictions only measured in the
Niranjan removing null values and fields covid 19 based not on the terms of R-
which are non-significant for on regression. overall analysis Squared(R2 ),
YEAR - 2020 this study. In the training stage, 2. The LR model is of the dataset. mean squared
the model is trained and tested known to error(MSE) and
handle quite a mean absolute
JOURNAL – for prediction. The results are
large data set error(MAE)
evaluated using measures like
International Journal of MSE, MAE.[4] and is hence
Scientific & applied giving a very
Science good
[Link] performance of performance in
prediction is compared between this paper
PUBLISHER- Linear Regression which is a 3. The major
IEEE commonly used type of disadvantage
predictive analysis and is used of SVR(Support
to foresee a numeric result Vector Regression)
given an arrangement of is that it cannot
autonomous factors and handle large data
Support Vector Regression. It is sets and hence its
a common application form of performance is
Support Vector Machine that less compared to
supports linear and non-linear LR(Linear 1
C. [Link].
Raman Global University, Regression)
Odisha
Problem Statement
COVID-19 had a devastating impact on economic and social development and people’s life
all over the world.
In the advent of covid-19 its inevitable to predict the growth trends in order to contain the
outburst, so a fast and efficient model which is capable of predicting the growth of covid-19
is need of the hour.
Due to the false entries of COVID, it shows positive cases in datasets.
Our goal from this task is to build a model that predicts the spread of the virus in the next
upcoming days with accurate results.
Fig4: Represents the germs of Corona Fig5: Represents how the germs gets1 double of Corona Virus
Virus
C. V. Raman Global University, Odisha
Methodology
Collection of
Datasets
Fig6: Represents the work flow
10
Methodology
Dataset collection:-
A data set is a collection of related, discrete items of related data that may be
accessed individually or in combination or managed as a whole entity. A data
set is organized into some type of data structure.
Fig 7:Represents the Dataset for prediction
1
C. V. Raman Global University, Odisha
Methodology
Data Preprocessing:-
The process of preparing the raw data and making it suitable for a machine learning
model. It is the first and crucial step while creating a machine learning model.[5]
The various data preprocessing techniques are as follows:-
o Data Manipulation
o Data Visualization
Fig8: Represents Data Manipulation for
finding missing values from the data Fig10:Represents the plot for all the
Fig9:Represents filling of Null
Values in Rows And Column states for the covid cases in India
12
Methodology
Data Preprocessing:-
In order to train a ML Model we have used different attributes of the dataset.
[6]
The Different Attributes are like Num cases ( gives us the information
about the total active cases present in that state ) , Current Status( gives us
the information about the current stage of the people who are hospitalized ),
Day and Month (gives us the information about the fact how it is varied ).
Fig12:Represents the Attributes used in model Training
Fig11:Represents the Attributes of the Dataset Used for
model Training 13
Methodology
Model Training:-
The Random Forest Algorithm is used for training a model.
Random forests or random decision forests is an ensembles
learning method for classification, regression and other tasks that operates by
constructing a multitude of decision tree at training time.
Scikit-learn (Sklearn) is the most useful and robust library for machine
learning in Python. [7]
It provides a selection of efficient tools for machine learning and statistical
modeling including classification, regression, clustering and dimensionality
reduction via a consistence interface in Python.
Fig13: Represents The evaluation of Models
14
Methodolog
y
Model Evaluation:-
Model evaluation is the process of using different evaluation metrics to understand a
machine learning model's performance, as well as its strengths and weaknesses.
Model evaluation is important to assess the accuracy of a model .
Fig14: Represents The evaluation of Models
15
Results States wise prediction
Fig15: Maharashtra state prediction for confirmed cases Fig16: Tamil Nadu state prediction for confirmed cases
Fig17: Kerala state prediction for confirmed cases Fig18: Telangana state prediction for confirmed cases 16
Results
Fig1: Prediction using Linear Regression Fig2: Prediction using Polynomial Regression
Accuracy- 2.171 Accuracy- 97.91
Fig3: Prediction using Random forest
17
Accuracy-98.63
Future-Work
After the model evaluation and accuracy we need to go for deployment of the ml model in
web app.
Fig19:Representing Future Works for this project
18
Conclusion
In this project it is concluded with a methodology and strategy for predicting
the COVID-19 outbreak.
Furthermore, it is possible to forecast the pandemic trend using data
collected from other perspectives.
We have also visualized and forecast the trend of covid-19 in the most
affected zones of India ( Telangana, Maharashtra, Kerala and Tamil
Nadu)
After predicting from different algorithms, it is concluded that Random
forest shows better accuracy among tested algorithms.
19
References
[1] R. Zunino and S. Rovetta R, “Covid-19 Future Forecasting ,” IEEE Trans. Ind. Electron. , 47 159 –167 (2000).
[2] T. Naito, “Covid-19 Future forecasting using support vector machine ,” Proc., IEEE/IEEJ/JSAI Int.( 2015)
[3] H. A. Hegt, R. J. Haye, and N. A. Khan, “A high performance Covid-19 Predictions,” Proc. IEEE International Conference on Systems, Man, and
Cybernetics, pp. 4357–4362 (1999).
[4] T. Naito et al.;, “Robust Machine Learning Performance,” IEEE Trans. Veh. Technol. , (1999)
[5] [Link], “Covid-19 Occurrences in Wuhan China ," International Journal of Engineering and Technology Vol.2
(3), 200-206, 2010.
[6] Saeed Rastegar, Reza Ghaderi, Gholamreza Ardeshipr & Nima Asadi, " An intelligent control system using an
efficient Algorithm for Building Approach", International Journal of Image Processing (IJIP) Volume(3), 2009.
[7] Wisam Al Faqheri and Syamsiah Mashohor, "A Real-Time comparision between the Decision tree and Random
Forest Algorithm“ ,IJCSNS International Journal of Computer Science and Network Security, (2003)
[8] Satadal Saha1, Subhadip Basu, Mita Nasipuri, Dipak Kumar Basu, ”Machine Learning Approach for prediction
forecast of covid-19 pandemic”, International Journal of Recent Trends in Engineering, Vol 1, No. 1, May 2009.
1
C. V. Raman Global University, Odisha
Thank You