



































East


East Afr. J. Biophys. Comput. Sci. (2022), Vol. 3, Issue. 1, 58-68 
East 

 

 

 

*Corresponding author: 

Email: beshow.betela@gmail.com, +251 972691341 https://dx.doi.org/10.4314/eajbcs.v3i1.6S 
 

 

 

Analysing COVID-19 Verified, Recuperate and Death Cases in Ethiopia Using ARIMA Models 

 

Birhanu Betela Warssamo 

 

Department of Statistics, College of Natural and Computational Sciences, Hawassa University, 

Hawassa, Ethiopia, P. O. Box: 05 

 

 

KEYWORDS:  

COVID-19 Cases; 

Days; 

Autoregressive Integrated 

Moving Average 

 

 

 

 

 

 

 

 

 

 

 

ABSTRACT 

Applying a successful prediction of the confirmed, recovered and deaths is supposed to be 

the basic requirement to successfully control the dissemination rate of diseases. Time 

series models have widely been considered as the suitable methods to forecast the 

confirmed, recovered and deaths because of the virus. The objective of this research is to 

apply the Autoregressive Integrated Moving Average (ARIMA) modelling approach for 

projecting COVID-19 confirmed, recovered and deaths cases in Ethiopia. Over strict 

tracks of all phase of Box-Jenkins strategy, ARIMA (1,1,1), ARIMA (16,1,2) and ARIMA 

(0,1,1) models for confirmed case, recovered and death case, respectively were selected as 

the most appropriate models for predicting corona virus cases of Ethiopia. Using these 

models, a predictions of five month a heads future situation of COVID-19 confirmed case, 

recovered and death case (Jan 3, 2021 to May 3, 2021) has made. The results showed that 

in the coming five months from Jan 3, 2020 to May 3, 2021, the number of COVID-19 

confirmed, recuperated and deaths cases in Ethiopia may reach up to 320,597; 168,912 

and 4438, respectively. By and large, the size of the corona virus distribution was 

increased from time to time in the past ten month, until 3rd Jan, 2021, and it is anticipated 

to continue faster than before for the coming 5-months, until the end of May, 2021, in 

Ethiopia and more rapidly than before while the peak will remain unknown yet. Therefore, 

an effective execution of the precautionary measures and a rigorous compliance by 

circumventing carelessness with the rules such as prohibiting public congregations, travel 

restrictions, social distancing, and strict use of personal protection measures may improve 

the dissemination rates of the virus. Further, through updating more new data with 

constant reconsideration of predictive model provide useful and more accurate prediction 
 

 

INTRODUCTION 

The primary event of corona virus (COVID-19) 

eruption was informed on December 31, 2019 in 

Wuhan and then, it has been put across the 

universal deadly disease owning harsh new type 

of hazard to individual vigour and life in a 

continues way. According to World Health 

Organization (2020), more than 135,045 

COVID-19 cases, 2,083 deaths, and 121,594 

recoveries have been reported in Ethiopia and 

101,483,628 COCID-19 cases, 2,185,413 deaths 

and 73,386,329 recoveries and 25,908,918 

active cases among these 25,801,597(99.6%) in 

East African Journal of Biophysical and Computational Sciences 

Journal homepage : https://journals.hu.edu.et/hu-journals/index.php/eajbcs 
 

 
Hawassa University

College of Natural & Computational Sciences

Year 2021

Volume xx No xx

 
Research article

file:///D:/EAJBCS/Birhanu%20Betela/bes
https://dx.doi.org/10.4314/eajbcs.v3i1.6S


East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

59 

mild condition and 110,289(0.4%) series or 

critical cases and 73,386,329 (97%) are 

recovered and 2,185,413(3%) are dead across 

the globe during the study period (WHO, 2020). 

The quantity of COVID-19 cases rapidly goes 

up in mid-January, and the infection quickly 

blowout beyond China’s boundaries. As of this 

research, it has extended to 219 countries all 

over the globe (WHO, 2019). Despite 

broadening COVID-19 influences, there are key 

public health questions. How many individuals 

will be diseased? How does the condition vary 

day by day? How many individuals will die? 

These inquiries could be re-joined by predicting 

the possible futures of this contagion via time 

series models. However, broadly used time 

series models and tools do not essentially have 

high forecasting precision, particularly for 

medical researches (Huppert and Katriel, 2013). 

By fitting models and creating precision 

forecasting unfailingly may improve community 

health judgment-making (Yonar et al., 2020). 

On March 13, 2020, the Federal Ministry of 

Health has confirmed a corona virus disease 

(COVID-19) index case in Addis Ababa, 

Ethiopia. The case, which was publicized on the 

13th of March 2020, is the first one (index case) 

to be reported in Ethiopia since the beginning of 

the outbreak in China in December 2019. The 

case is a 48- year old Japanese man reported to 

have travelled from Japan to Burkina Faso and 

then reached in Ethiopia. The person developed 

symptoms and presented at the health center in 

Addis Ababa from where the rapid response 

team (RRT) moved him to the isolation facility 

in Yeka Kotebe. He was clinically stable, with 

no serious symptoms. Subsequently, on 16 

March 2020, the administration of  Ethiopia 

have been announcing to appliance numerous 

defensive actions such as emerging community 

alertness about over all behavior of the infection 

and its stoppage tools like cleansing of ways and 

marketplaces, isolation of mistrusted and 

diseased cases, lockdown of the universities, 

schools, house of worship, and announcing state 

of crisis. Ethiopian administration has applied 

protective measures directly after the infection 

has been stated the world disease. Yet, the 

spread has continued quick for the reason that 

supreme of the individuals have been 

carelessness in certain of defensive actions such 

as public distancing and mass assembly. On the 

other, populace movement through the border 

from the next-door nations like Sudan, Djibouti, 

Kenya and Somalia has a lion part for the 

current movement of quick spread of infection 

in Ethiopia. Accordingly, COVID-19 spread 

would get thoughtful in Ethiopia, and its coming 

movement is also likely to depend directly or 

indirectly on the spread of COVID-19 within 

and outside of all those neighboring African 

countries. 

Investigating the trends and predicting the 

coming indices of the infection in Ethiopia must 

be accompanied in order to put in a place 

operative monitoring policies. Though the 

spread mechanisms of the infection is not totally 

recognized, the number of new cases are harshly 

growing, and the effects of inhibition should be 

assessed on numerical data where numerical 

investigation is extra significant. 

Actually, several states in the world up to now 

have functional data-driven statistical models 

alike Autoregressive Integrated Moving 

Average (ARIMA) for forecasting of future 

movements of the communicable infection. 

Subsequently, to model the exponential growing 

rate of the virus, Yichi et al. (2020) and Gupta  

and Pal (2020) conducted their study using a 



East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

60 

linear based time series model, Auto Regressive 

Moving Average (ARIMA family) model. Time 

series forecasting models like ARIMA is the 

ordinary method which gives clothed forecasts 

on time series data in fast time (Hyndman et al, 

2008). This method of investigation has been 

extensively functional, for its consistency and 

rapid application by numerous stakeholders. 

Hence the objective of this work is to put on the 

Autoregressive Integrated Moving Average 

(ARIMA) modelling approach for predicting 

corona virus (COVID-19) confirmed cases, 

recovered and deaths in Ethiopia to empower 

the community health organizations to conduct 

trust worthy day-to-day prediction and to arise 

with appropriate intervention policies, deliver 

direct and long-term route of the illness and play 

role to the body of knowledge in epidemiologic 

study method. 

MATERIALS AND METHODS 

Study Area 

The research was carried out in Ethiopia, where 

the virus quickly spread and disturbing. 

According to the UN population estimates, 

whole population of the country is 114,963,836 

peoples. 

Research data  

The everyday time series data of COVID-19 of 

Ethiopia from 13 March 2020 to January 3, 

2021 were gathered from the official website of 

Johns Hopkins university (2020): 

https://goto.now/AWY0o. SPSS Version 20 

statistical software was functional to achieve 

statistical data investigation on the confirmed, 

recovered and deaths case of COVID-19 

datasets. 

Technique of data study 

Time series analysis aims to tell consistent and 

expressive statistics and use this information to 

forecast upcoming values of the series. Time 

series models try to predict the upcoming values 

by investigating the former and present. It 

contemplates past data and tries to arise some 

procedure which will clarify those existences 

and forecast upcoming values. The distinct 

feature of time series analysis is consecutive 

observations are typically dependent and that 

the investigation must take methods to classify 

the designs which characteristically happen in 

the data. Based on this detail, amongst time 

series methods, this research applied ARIMA 

model in order to evaluate the upcoming 

movement situation of COVID-19 in Ethiopia.  

Auto Regressive Integrated Moving Average 

(ARIMA) modelling 

It is the other most regularly used time series 

models as it considers varying movements, 

periodic fluctuations and random disturbances 

in the time series. It also appropriate for all 

types of data, containing non-stationary data, 

which is if there is no systematic change in 

mean (no trend), no systematic change in 

variance and periodic variations has removed 

(Box and Jenkins, 1976).  In practice, most of 

the time series are non-stationary and Wei, 2006 

commends eliminating any non-stationary 

sources of variation in time series data. In most 

case, common method for attaining stationary is 

to put on consistent differencing and log 

transformation to the original time series (xt). If 

differencing a time series d times results in a 

stationary series, then that original series is said 

to follow an Autoregressive Integrated Moving 

https://goto.now/AWY0o


East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

61 

Average (ARIMA) process, denoted as ARIMA 

(p, d, q) and can be written as: 

    tt

d WX  
 

Where: 

Autoregressive operator can be expressed as: 

  p

p  ...1 2

2  

Moving average operator 

   p ...1 1
 

Differencing operator    dd  1  it is the 

expression of dth consecutive differencing so as 

to make series stationary. Wt is a Gaussian white 

noise series with mean zero and variance  2

w .
 

Testing for stationary: Before emerging a 

Box-Jenkins modelling process, it is significant 

to verify whether the data under study 

encounters basic suppositions such as series 

stationary. A time series is well-thought-out as 

stationary if its statistical properties such as 

mean and variance are constant over time (Box 

and Jenkins, 1976).  Many testing procedures 

for stationary are planned in the literature. In 

this research, correlogram test were functional 

for testing whether the series is stationary. 

The correlogram test: It is one way to 

characterize a series with respect to its reliance 

over time. It usually recognized as sample 

autocorrelation function (ACF), which is plot of 

sample ACF coefficient against observation 

difference in time (lag). As indicated in (Box 

and Jenkins, 1976) if the sample ACF decays 

very slowly in non-seasonal and seasonal lag 

snit indicates that differencing is wanted and an 

insinuation for series non-stationary. 

Constructing ARIMA model: As planned by 

(Granger and Newbold, 1986) method, in order 

to construct ARIMA model for a specific time 

series data, must follow four phases: Model 

identification, estimation of model parameters, 

Diagnostic checking for the identified model, 

Application of the model (forecasting). 

Model Identification: With this phase, the 

amount of differencing compulsory attaining 

stationary and the order of both the seasonal and 

non-seasonal AR and MA operators are 

determined. The autocorrelations function 

(ACF) and the partial autocorrelation functions 

(PACF) are the two most valuable tools in any 

attempt at time series model identification 

(Shumway and Stoffer, 2010). To determine the 

number of differencing (d), non-seasonal 

autoregressive (p) and moving average (q) 

parameters, the guideline specified in (Lehmann 

and Rode, 2001) are used. Therefore, if PACF 

Cuts off after lags q and ACF tail off, then 

ARIMA (0, d, q) model is identified. If ACF cut 

off after lag p and PACF tail off, ARIMA (p, d, 

0) model is acquired. Finally, if both ACF and 

PACF tail off, then the recognized model will 

be ARIMA (p, d, q). 

Parameter Estimation: After selecting the 

most suitable ARIMA model, the parameters are 

estimated by Maximum Likelihood Estimation. 

Diagnostic Checking: it deals with the residual 

assumptions in order to decide whether the 

residuals from fitted model are independent, 

constant variance (Lehmann and Rode, 2001). 



East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

62 

Visual Analysis: It is a method for investigating 

plot of the residual over time. If visual reviews 

of the plot tell that they are haphazardly 

distributed over time, then it is a residual 

independence (Lehmann and  Rode, 2001). 

Residual Autocorrelation Function (RACF): 

With this test, to say that residual follows a 

white noise process, roughly 95% of the 

autocorrelation coefficient should fall within the 

range of ±1.96/squ(n) (Grasselli et al, 2020). 

 If the nominated model is insufficient, the 

three-step model building process with other 

model is typically repeated many times until a 

satisfactory model obtained. The final model 

selected can then be used for forecast 

determinations (Box and Jenkins, 1976). 

 
Figure 1 Distribution of confirmed cases of covid-

19 in Ethiopia(Mar 3, 2020- Jan 3, 2021) 

Predicting: It is last phase in time series 

modelling, the aim is to forecast future values of 

a time series, 𝑋𝑡 + 𝑚, 𝑚 = 1,2,… based on the 

data collected to the present, 𝑋 = {𝑋𝑡 , 𝑋𝑡 −

1, … , 𝑋𝑙}. 

qmtqmtdpmtdpmtmt WWXXX    ...... 1111  

RESULTS 

Descriptive analysis results 

The first confirmed case of the COVID-19 in 

Ethiopia (the index case) is described on 13 

March 2020 with one infected foreigner. Then 

after, the administration of Ethiopia has without 

delay in use different measures to put off and 

manage the infection, but the case has been 

expontially increasing and widely distributing 

until Sep 1, 2020. From Aug 01, 2020 to Oct 01, 

2020 large number of people are infected by the 

virus and the number of infection shows slight 

decline until the study period Jan 3, 2021 

(Fig.1). Similarly, the first recovered cases 

reported on Jun 01, 2020 and the recovery cases 

increased parallel with the confirmed cases. 

There was a large number recovery cases during 

December 1, 2020 to January 1, 2021 (Fig 2).  

 
Figure2: Distribution of covid-19 recovered case 

in Ethiopia (13 Mar, 2020-Jan 3, 2021) 

As shown in Fig 3, the first mortality/death case 

was reported on April 2, 2020. The number of 

deaths was increasing with the increase of the 

number of confirmed cases. High number of 

deaths in Ethiopia was reported during Aug 03, 

2020 – Oct 1, 2020. As shown on Fig 4, the 



East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

63 

confirmed numbers of cases are increasing and 

the recovery case is also more increasing 

parallel to the confirmed cases. 

 

 

Figure 3: Distribution of death cases in Ethiopia 

(Mar 3, 2020-Jan3, 2021) 

The number of death cases is also increasing but 

interestingly the recovery cases (green line) is 

very closed to confirmed cases (blue line) but 

the death (yellow line) is going away from the 

confirmed cases (blue line), that is, the death 

cases are not increasing in the same level as 

confirmed cases and the death line is getting 

flat.  

 

Figure 4: Covid-19 confirmed, recovered and 

death cases in Ethiopia (Mar 3, 2020-Jan3, 2021). 

 

 

Results from ARIMA modelling of COVID-

19 case 

Result in testing stationary: The first step in 

every time series investigation containing 

ARIMA modelling is to see whether the time 

series is stationary. In this research, time series 

plot and Autocorrelation Function (ACF) were 

practical to check the stationary. The pattern of 

the series of COVID-19 confirmed cases, 

recovery and death cases shows increasing trend 

indicating the mean and the variance of the 

series are not constant throughout time and this 

shows the process is not stationary and the data 

need to be differencing and with the one step 

differencing, d=1, we got stationary data.  

Result in model identification for confirmed 

cases:  

From Table 1 ARIMA (1,1,1) and ARIMA 

(5,1,1) are recommended since they have lest 

Bayesian Information Criterion (BIC) and 

highest adjusted R-square, but by B-J 

methodology, parsimonious models give better 

 

Transforms: natural log      Confirmed case_mean 

     Recoverd_mean 

     Deathcase_sum 



East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

64 

 

forecast than over-parameterized models. 

Therefore, ARIMA (1,1,1) is selected. 

 

Table 1: Tentative models 

Confirmed cases ARIMA(1,1,1) ARIMA(1,1,2) ARIMA(5,1,1) ARIMA(5,1,2)   

Adj R square 88.9 88.9 89.3 89.3 

BIC 9.855 9.877 9.911 9.933 

 

Result in model identification for recovered 

case: 

From Table2 ARIMA (15,1,1) and 

ARIMA(1,1,2) are selected but by B-J 

methodology , ARIMA (1,1,2) is the best model 

to forecast the confirmed cases. But residual 

ACF and PACF are not flat, that is lags 16 and 

15 are very significant indicating there is white 

noise in the series and we should re estimate the 

model by adding AR(16) in the model. Thus the 

appropriate model to forecast recovery case is 

ARIMA (16, 1, 2).  

 

Table 2: Tentative models 

Recovered  cases ARIMA(1,1,1) ARIMA(1,1,2) ARIMA(15,1,1) ARIMA(15,1,2) 

Adjusted R square 60.4 60.6 66.1 66.4 

BIC 11.614 11.631 11.775 11.788 

 

Result in model identification for death case:  

From auto and partial auto correlation plot a 

candidate MA(q) models were recommended. 

That is ARIMA (0,1,1) models is recommended. 

Result in model diagnosis: According to (Box 

and Jenkins, 1976), all the models that accepted 

and satisfied all residual tests and the 

parameters meaningfully vary from zero must 

be included and nominated as candidate model 

for forecast. Residuals ACF and partial ACF 

plot for confirmed, recovered and death cases, 

all the lags are inside the 95% confidence 

interval showing there is no problem of white 

nose so the models are ready for use of 

prediction. 

 

 

 

 

 



East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

65 

 

Table 3: ARIMA models & coefficients for best selected model. 

COVID-19 cases Model BIC R-squared MAPE Coefficient P value 

Confirmed case ARIMA(1,1,1) 9.855 .889 0.309 
1



 =-0.233 

0.016 

     
417.0



  
0.00 

Recovered case  ARIMA(16,1,2) 11.732 0.69 0.439 
1



  =0.257 
 

0.00 

2



  =0.230 

0.00 

3




 =0.129 

0.034 

6




 =0.185 

0.003 

15




 =-0.330 

0.00 

998.0



 

0.00 

Deaths cases ARIMA(0,1,1) 2.649 0.679 0.407 
801.0



  
0.00 

 

Predicting precision assessment 

If the fitted models achieve well in predicting, 

the estimate error will be comparatively minor 

and the Mean Absolute Percentage Error 

(MAPE) should be close to 5%. From Table 3, it 

can be witnessed that the accuracy of forecasts 

measured by the Mean Absolute Percentage 

Error (MAPE) cast out to be 3.09% for ARIMA 

(1, 1, 1) model, 4.39% for ARIMA (16, 1, 2) 

and 4.07% for ARIMA (0, 1, 1) model, which 

are relatively less than 5%. This implies that 

those models would perform better in predicting 

the confirmed, recovery and death cases of 

COVID-19 well.  

 

 
 

Figure 5: Forecasted value with 95% prediction 

interval for confirmed cases in Ethiopia 

Figure 6: Forecasted value with 95% prediction 

interval for recovered cases in Ethiopia 

 



East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

66 

 

Fig 5, Fig 6 and Fig 7 presents prediction of 

COVID-19 confirmed, recovered and death 

cases and it gives the direction and tendency of 

the epidemic and forecasts the likely 

development of future epidemics. The Figs 

shows, a forecasts of five month a heads future 

situation of COVID-19 confirmed case, 

recovered and death case (Jan 3, 2021 to May 3, 

2021). Infected cases, recovered and death cases 

are expected to considerably increase in the 

coming five months. By the end of May 3, 2021, 

cumulative confirmed, recovered and death 

cases across Ethiopia, will reach 320,597, 

168,912 and 4438 respectively. 

 

 

Figure 7: Forecasting value with 95% prediction interval for deaths case in Ethiopia 

 

DISCUSSION 

It is crucial to generate consistent and 

appropriate forecasting model that can 

assistance health system administrations and 

other stakeholders to prevent the extra blowout 

of COVID-19. Time series forecasting models is 

the statistical method which provides a clothed 

forecast and has been extensively practical for 

trend of communicable sickness in rapid time 

(Fanelli and Piazza 2020). ARIMA models are 

predictive technique that offers a good forecast 

and has been widely used for the rapid trend of 

infectious diseases (Holt, 1957, Hyndman et al, 

2008, Shumway and Stoffer, 2010). 

Over strict follows of all phase of Box-Jenkins 

strategy, ARIMA (1,1,1), ARIMA (16,1,2) and 

ARIMA (0,1,1) models for confirmed case, 

recovered and death case respectively were 

selected as the best models for predicting 

COVID-19 cases of Ethiopia. Time series 

analysis of COVID-19 provides the direction 

and movement of the epidemic and forecasts the 

probable progress of upcoming epidemics. 

Using these models, a forecasts of five month a 

heads future situation of COVID-19 confirmed 

case, recovered and death case (Jan 3, 2021 to 

May 3, 2021) has made. In diverse area, a lot of 

numeric researches have been showed to 

forecast new case of COVID-19 case using 

ARIMA model (Gupta and Pal, 2020; Li et al., 



East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

67 

 

2020, Yonar et al., 2020; Zhan et al., 2020) 

which are consistent with this research. In line 

with the current study, Li et al. (2020) and Das 

(2020) also used ARIMA model to predicting 

COVID-19 epidemic in India and China, 

respectively and showed that the infected case 

were increasing. 

COVID-19 cases reported in the country is 

assumed to exceed 160,585 in the five months 

of 2021. At the end of May, inline projected 

cumulative infections across Ethiopia to hit 

156,610 on average (Hyndman et al, 2008), 

while the real value until 19 May was just 

189,137 cases and 1352 deaths (Yonar et al., 

2020). This means that the number of people 

who are corona-virus-positive in Ethiopia can 

increase more than predicted (Hyndman et al, 

2008). However, in this analysis, the cumulative 

confirmed, cumulative recovered and 

cumulative death forecast cases on May 3, 2021, 

are 320,597, 168,912 and 4438, respectively. 

Similar to the predicted values in this study, 

variation of pandemics in Ethiopia in the first 

five months has been shown to increase 

(Hyndman et al, 2008). The scholars disprove 

the effect of geographical difference and 

temperature in reducing the distribution of 

COVID-19 (Granger and Newbold, 1986). The 

COVID-19 pandemic has caused significant 

global social and economic disruption. While 

pre-protective strategies are crucial in managing 

the spread of COVID-19, this study does not 

evaluate their impact due to the unavailability of 

exposure data 

CONCLUSIONS 

Since the study shows the increase in the 

number of deaths and confirmed cases in 

Ethiopia, more attention should be given to the 

control and prevention of Covid-19. Without the 

implementation of effective infection control 

measures by the Government of Ethiopia, the 

COVID-19 situation is expected to deteriorate, 

leading to greater consequences for the nation. 

Consequently, this study advocates for the 

development and enforcement of sustainable 

pandemic control strategies. Utilizing the 

ARIMA model allows for forecasting the future 

spread of COVID-19 based on existing data, 

which will assist institutions in formulating 

appropriate policies.  

 

References 

Box G. and Jenkins G. 1976. Time series analysis: 

Forecasting and control. Holden day. 

Canters for Disease Control and Prevention. Corona virus 

Disease 2019 (COVID-2019).  [https://www. 

cdc.gov/coronavirus/2019-ncov]. Accessed on 1-28-

2021. 

Das R. 2020.  Forecasting incidences of COVID-19 using 

Box-Jenkins method for the period July 12-

Septembert 11, 2020: A study on highly affected 

countries. Chaos, Solitons and Fractals 140: 110248. 

Link: https://bit.ly/2MBS4XR 
Domenico B., Marta G., Lazzaro V., Silvia A., and, 

Massimo C. 2020. Application of the ARIMA model 

on the COVID- 2019 epidemic data set. Data Brief 

29: 105340. 

Fanelli D. and Piazza F. 2020. Analysis and forecast of 

COVID-19 spreading in China, Italy and France. 

Chaos, Solitons& Fractals, 134, 

1e12https://doi.org/10.1016/j.chaos.2020.109761 

 Fattorini D. and Regoli F. 2020.  Role of the chronic air 

pollution levels in the Covid-19 outbreak risk in 

Italy. Environ Pollu 264: 114732.  

Granger K. and Newbold J. 1986. Forecasting economic 

time series. USA: Academic Press. 

Grasselli G., Pesenti A. and Cecconi M. 2020. Critical 

care utilization for the COVID-19 outbreak in 

lombardy, Italy: Early experience and forecast 

during an emergency response. J. Am. Med. Asso. 

323(16): 1545-1546. 

Gupta R and Pal S.K.. 2020. Trend analysis and 

forecasting of COVID-19 outbreak in India. 

MedRxiv. 

https://doi.org/10.1101/2020.03.26.2004451. 

Holt CE. 1957. Forecasting seasonal and trends by 

exponentially weighted averages (O.N.R. 

https://bit.ly/2MBS4XR
https://doi.org/10.1101/2020.03.26.2004451


East Afr. J.Biophys.Comput. Sci. (2022), Vol. 3, No. 1, 58-68 
 

68 

 

Memorandum No. 52). Carnegie Institute of 

Technology, Pittsburgh USA. 

https://doi.org/10.1016/j.ijforecast.  

Huppert A and Katriel G. 2013. Mathematical modelling 

and prediction in infectious disease epidemiology. 

ClinMicrobiol Infect. 19(11):999-1005. 

https://doi.org/10.1111/1469-0691.12308 PMID: 

24266045 

Hyndman R.J., Koehler A.B., Ord J.K., and Snyder R.D. 

2008. Forecasting with Exponential Smoothing: The 

State Space Approach. Berlin Germany: Springer, 

372 p. 

Khana F. and Gupta R. 2020. ARIMA and NAR based 

prediction model for time series analysis of COVID-

19 cases in India. Journal of Safety Science and 

Resilience 1: 12-18. Link: https://bit.ly/3ot9YtC 

Lehmann A.  and  Rode  M. 2001. Long-term behaviour 

and cross-Correlation Water quality analysis of the 

river elbe, Germany. Water Res. 35: 2153-2160. 

Li Q., Feng W. and Quan Y. 2020. Trend and forecasting 

of the COVID-19 outbreak in China. J Infect 80 (4): 

469-496. https://doi.org/10.1016/j.jinf.2020.02.014 

Papastefanopoulos V., Linardatos P.,  and, Kotsiantis S. 

2020. A Comparison of Time Series Methods to 

Forecast Percentage of Active Cases per Population. 

Applied sciences 10: 3880. Link: 

https://bit.ly/35bxKTy 

Shumway  R. and Stoffer D. 2010. Time series analysis 

and its applications with R Examples (3rd ed.). 

Springer 

Wei W.  2006. Time series analysis univariate and 

multivariate (p. 478). New York-USA: Addison-

Wesley Publishing Company, Inc. 

WHO. Corona virus disease 2019 (COVID-19) situation    

reports. [https://www.who.int/emergencies/ 

diseases/novel-coronavirus-2019/situation-reports] 

Accessed: 2020-05-05. 

World Health Organization (WHO) 2020. Corona virus 

disease (COVID-19) pandemic, WHO  

https://worldometer.info/coronavirus.Assesedon.Jan 

28, 2021. 09:09  

Worldometer, “Coronavirus Cases,” Worldo meter, pp. 1–

22, 2020, doi: 1101/2020.01.23.20018549V2. 

Assessed on Jan 28, 2021. 09:09 GMT 

Yichi L., Bowen W., Ruiyang P., Chen Z., Yonglong Z., 

Zhuoxun L, Xia J. And, Bin Z. 2020. Mathematical 

Modelling and Epidemic Prediction of COVID-19 

and Its Significance to Epidemic Prevention and 

Control Measures. Annals of Infectious Disease and 

Epidemiology 5 (1): 1052. 

Yonar H., Yonar A, Tekindal M.A and, Tekindal M. 

2020. Modelling and Forecasting for thenumber of 

cases of the COVID-19 pandemic with the Curve 

Estimation Models, the Box-Jenkins and 

Exponential Smoothing Methods. Eurasian J. Med. 

Oncol. 4(2):160–165. 

Zhan C., Chi K., Lai Z., Hao T., and, Su J. 2020. 

Prediction of COVID-19 spreading profiles in South 

Korea, Italy and Iran by data-driven coding. 

medRxiv.. 
  
 

https://doi.org/10.1016/j.ijforecast.%202003.09.015
https://bit.ly/3ot9YtC
https://doi.org/10.1016/j.jinf.2020.02.014
https://bit.ly/35bxKTy
https://worldometer.info/coronavirus.%20Assesed%20on.Jan

