Data based Analysis, Modeling and Forecasting of Novel Coronavirus in infected regions using Extreme Learning Machine algorithm

In current development in science and technology, Machine learning algorithms play an essential role for prediction, classification, data analysis and data visualization. With this efficient algorithm, we can solve many real-world problems in all domains like education, healthcare, banking, geographical analysis, etc., in the current scenario; much research work is going on with the new virus's infection called the corona. This Corona virus is a comprehensive unit of virus this cause illness in humans or animals, now in East Asian countries, this virus affected more people. In India, the first case was found in January month, originated from China. The entire world is focusing on the disease, and day by day, the infection and death rate is increasing. In this, we intended to focus on the spread of this deadly disease and to demonstrate which countries are the most affected by doing statistical analysis. On December 2019, As of 10 February 2020, China reported overall of 40,235 cases 909 deaths, evoking local and foreign terror. Here we provide estimates of the major epidemiological parameters, Based on the epidemiological data available to the public for Hubei, China, 11 January to 10 February 2020. In particular, we give an estimate Fatality and case recovery rates, along with their 90 per cent confidence levels as the epidemic progresses. For this work implementation, Extreme Learning Machine algorithm used. ELM is a feed-forward network and its learning rate also fast when compare to normal neural network. No need to provide any weights and bias values. This algorithm will give a promising result with the best accuracy.


Introduction:
Corona-viruses are a wide range of viruses which can source disease in natural world or individual. Corona viruses' symptoms are common cold to severe diseases such as respiratory syndrome is well-known to cause human respiratory infections. Novel Corona-virus discovered causes a COVID19 Corona-virus disease. Many individuals who have the infection get COVID-19. The bad health extend principally From individual to person, by tiny nose or mouth beads that are stripped off when a person with COVID-19 hacks, sniffles or talks. Individuals get this virus infection if they take these beads in. So we have to stay away from others to stop spreading this virus. Such beads for example surface like tables and objects like door hndles which are touched by infected person. Person may get contaminated by rubbing their eyes, nose, or mouth after touching the objects or surface which are used by infected person . World Health Organization advised to wash our hands 2 for minimum 20 minutes with soap and water. [2] Six modes are proposed by UCCSF, this is self preservation mode, COVID containment mode, economic survival mode, herd immunity mode and Darwinian mode. It has traversed approximately 180,000 confirmed infections worldwide as of March 17, 2020, including 7426 deaths. On 11 March the WHO declared it to be a pandemic. The size of the outbreak has exceeded diseases caused by two other significant corona viruses to date. The overall objective of this work is to focus on the spread of the deadly disease. To demonstrate which countries are the most affected by doing statistical analysis. Machine learning algorithms play an essential role for prediction, classification, data analysis and data visualization. To achieve this proposed work Extreme Learning Machine algorithm ELM algorithm. 1.1 About ELM [1] Normal feed forward neural network has lac of fast learning method; to solve this problem ELM is proposed. It has the ability to set the hidden layer and randomly assigned input weights It needs a single iteration for doing the learning process. It overcomes the drawback of the feedforward neural network also improves the learning ability and overcome the Overfitting problem Figure:1 Yearly publication using ELM algorithm This Graph shows that from 2014 onwards research works for doing predictions and data analysis; this ELM shows excellent results of accuracy.

About Linear Regression
The simplest regression models comprised of a single response variable Y and a single predictor X. STATGRAPHICS will suit several functional types. They will list the models in decreasing Rsquared order. When outliers identified, resistant methods may be used instead of least squares to satisfy the models.

Multiple Regression
The multiple regression method suits a model with various predictor variables X1, X2, ... in relation to an answer variable Y The user can include all predictor variables in the match, or ask the program to pick a subset with only relevant predictors using stepwise regression. Around the same time, the Box-Cox approach and the Cochrane-Orcutt test can be used to tackle nonnormality To test our model, we'll use Root Mean Square Error (RMSE) and determination of coefficient (R2 score).
RMSE is square root of the remaining square sum total. Were ss is the total sum of errors if the mean of the observed value as the predicted value.

About Neural Network:
This neural network is composed of several layers. Each layer is made up of nodes where the computation will take place. A node combines the input, coefficients set and weights. The input weights are multiplied and then summed, the sum given to the node which is called as an activation function. By using Neural Network is used for many application developments. This proposed paper consists of several statistical data analysis to identify the spread of viruses throughout India and all over the world. For predicting along with ELM, Linear regression and Multiple regression. [3] Now, India is in critical condition due to drastic increases in the rate of infection and also death. Compared to China India, U.S.A and Italy are facing more difficult because of this virus infection, and all Asian countries have been panic about the rise in death, diseases, and also affecting all nations' economic growth. [4] The impact of this pandemic will last for a long time, affecting many human lives and affecting all global development activities, including technical, financial and investment activities. To handle the situation, our government giving guidance with the help of Public health programs which are an equally essential branch of the health system to protect communities' health and keep them involved in disease prevention ICMECE 2020 IOP Conf. Series: Materials Science and Engineering 993 (2020) 012104 IOP Publishing doi:10.1088/1757-899X/993/1/012104 4 and leading safe and prosperous lives. Even though the virus spread is increasing drastically, also increase the number of a confirmed case, both India and Tamilnadu. This research work was focusing on analysis which is places are most affected by this disease spread, checking the death rate by ding the statistical analysis. In previous work, by implementing the Arima Model, they predicted that the infection confirmed person in India would go to 27lakhs for India total population.

Dataset details
In this paper, two datasets consider for the analysis was taken from Kaggle. The first dataset consists of 40 states of virus infection details. It consists of 9 attributes which are test taken time, date, state. In that state, patients belong to India and foreign countries, no of death, no of confirmed cases and death rate. The second dataset consists of virus infection details for Tamilnadu. All these two datasets collected from January to June and total records are more than 3000.
From the COVID 19 all-state record dataset first we analyzed that what is the total confirmation infected people in a particular state.

Figure:3 Dataset Statistical analysis
From this Graph, the total confirmed cases, Maharashtra is the most affected place in India and Tamilnadu in second place. This research work is focusing on statistical analysis to demonstrate which are places are most affected by this virus. [5]