Covid-19-Time-Series-Modelling

#Time series modelling

library(lubridate) # Simplifies Date and Time manipulation library(forecast) # Analyzing and Forecasting

Create sample COVID-19 cases data

cases <- c(580, 7813, 28266, 59287, 75700,

•        87820, 95314, 126214, 218843, 471497,
  

•        936851, 1508725, 2072113)
  

Convert to time series object

covid_ts <- ts(cases,

•            start = decimal_date(ymd("2020-01-22")), 
  

•            frequency = 365.25 / 7)
  

#Check the ACF and PACF plots Acf(diff(covid_ts)) # ACF plot Pacf(diff(covid_ts)) # PACF plot

Plot the time series

plot.ts(covid_ts, xlab = "Weekly Data", ylab = "Total Positive Cases",

•     main = "COVID-19 Pandemic")
  

Create Simple moving average model

ma_model <- ma(covid_ts, order = 3, centre = FALSE) #Forecasting 5 steps ahead forecast_ma <- forecast(ma_model, h = 5) Warning message: In ets(object, model = "ZZN", lambda = lambda, biasadj = biasadj, : Missing values encountered. Using longest contiguous portion of time series print(forecast_ma) # display the forecast Point Forecast Lo 80 Hi 80 Lo 95 Hi 95 2020.287 1505843 -103131.6 3114817 -954871.2 3966557 2020.307 1505843 -1135420.3 4147106 -2533620.5 5545306 2020.326 1505843 -2290602.9 5302288 -4300319.8 7312005 2020.345 1505843 -3692060.0 6703746 -6443663.6 9455349 2020.364 1505843 -5450010.4 8461696 -9132216.6 12143902 #Simple Exponential smoothing ses_model <- HoltWinters(covid_ts, beta = FALSE, gamma = FALSE)

Double Exponential Smoothing (Holt's method)

des_model <- HoltWinters(covid_ts, gamma = FALSE) print(des_model) # results Holt-Winters exponential smoothing with trend and without seasonal component.

Call: HoltWinters(x = covid_ts, gamma = FALSE)

Smoothing parameters: alpha: 1 beta : 1 gamma: FALSE

Coefficients: [,1] a 2072113 b 563388

Triple Exponential Smoothing (Holt-Winters)

Automatically chooses the right exponential smoothing model

tes_model <- ets(covid_ts)

Forecast 5 steps ahead

forecast_tes <- forecast(tes_model, h = 5)

Plot

plot(forecast_tes)

Automatic ARIMA selection

auto_arima_model <- auto.arima(covid_ts)

Manual ARIMA specification,

#AR1, differencing and MA1 with MAximum likelihood estimate manual_arima <- Arima(covid_ts, order = c(1,1,1),method='ML') print(manual_arima) # Arima result Series: covid_ts ARIMA(1,1,1)

Coefficients: ar1 ma1 0.9601 1.0000 s.e. 0.0519 0.2592

sigma^2 = 2.912e+09: log likelihood = -149.88 AIC=305.75 AICc=308.75 BIC=307.21

Plot the time series

plot.ts(covid_ts,

•     xlab = "Weekly Data", 
  

•     ylab = "Total Positive Cases",
  

•     main = "COVID-19 Pandemic (Weekly Cases Over Time)")
  

Forecast 5 steps ahead

forecast_arima <- forecast(manual_arima, h = 5)

Plot ARIMA forecast

plot(forecast_arima,

•  main = "ARIMA(1,1,1) Forecast of COVID-19 Cases",
  

•  xlab = "Weekly Data",
  

•  ylab = "Predicted Cases")