### The program was written by Heng Liu(2106258) ### ASP Assignment task 2 ### This program processes and visualizes the COVID-19 data from Swansea and Wuhan ### January 2022 ### NB - You need to install the openxlsx, reshape2 and ggplot2 package before running this program library(openxlsx) library(reshape2) library(ggplot2) ### Function Definitions ### generateDescStats - generates the descriptive statistics # Parameters passed: vector v # Returns: a vector containing the descriptive statistics values generateDescStats<-function(v) { # Create DescStats to hold the statistics, ready to be included in the data frame vDescStats<-NULL # Calculating basic descriptive statistics - mean vDescStats<-c(vDescStats,mean(v,na.rm=TRUE)) # Calculate standard deviation vDescStats<-c(vDescStats,sd(v,na.rm=TRUE)) # Calculating basic descriptive statistics - median vDescStats<-c(vDescStats,median(v,na.rm=TRUE)) # Calculating more basic descriptive statistics - max, min vDescStats<-c(vDescStats,max(v,na.rm=TRUE)) vDescStats<-c(vDescStats,min(v,na.rm=TRUE)) # Calculate range vrange<-range(v,na.rm=TRUE) vDescStats<-c(vDescStats,vrange[2]-vrange[1]) # Return vDescStats return(vDescStats) } ### Main Script ### ### 1. Load data into a data frame covidData # Load data file # NB - # Because the original file has two sheets, it cannot be imported # So first read China's data and load it into covidDataChina # The second time to read UK's covid data, named covidDataUK # Then merge the two data.frames into coviddata covidDateChina<-read.xlsx("covid.xlsx",sheet = 1) covidDateUK<-read.xlsx("covid.xlsx",sheet = 2) covidDate<-rbind(covidDateChina,covidDateUK) ### 2. plot the number of cases and deaths in the two countries # Because the chart is plotted on a monthly basis, the data span 22 months # Create labels for twocountries data frame bymonth<-1:22 bymonth # Summary of all cases in China on a monthly basis chinatotalcases<-c(covidDate[10,5],covidDate[39,5],covidDate[70,5],covidDate[100,5], covidDate[131,5],covidDate[161,5],covidDate[192,5],covidDate[223,5], covidDate[253,5],covidDate[284,5],covidDate[314,5],covidDate[345,5], covidDate[376,5],covidDate[404,5],covidDate[435,5],covidDate[465,5], covidDate[496,5],covidDate[526,5],covidDate[557,5],covidDate[588,5], covidDate[618,5],covidDate[649,5]) # Summary of all deaths cases in China on a monthly basis chinatotaldeaths<-c(covidDate[10,8],covidDate[39,8],covidDate[70,8],covidDate[100,8], covidDate[131,8],covidDate[161,8],covidDate[192,8],covidDate[223,8], covidDate[253,8],covidDate[284,8],covidDate[314,8],covidDate[345,8], covidDate[376,8],covidDate[404,8],covidDate[435,8],covidDate[465,8], covidDate[496,8],covidDate[526,8],covidDate[557,8],covidDate[588,8], covidDate[618,8],covidDate[649,8]) # Summary of all cases in UK on a monthly basis uktotalcases<-c(covidDate[650,5],covidDate[679,5],covidDate[710,5],covidDate[740,5], covidDate[771,5],covidDate[801,5],covidDate[832,5],covidDate[863,5], covidDate[893,5],covidDate[924,5],covidDate[954,5],covidDate[985,5], covidDate[1016,5],covidDate[1044,5],covidDate[1075,5],covidDate[1105,5], covidDate[1136,5],covidDate[1166,5],covidDate[1197,5],covidDate[1228,5], covidDate[1258,5],covidDate[1289,5]) # Summary of all deaths cases in UK on a monthly basis uktotaldeaths<-c(covidDate[650,8],covidDate[679,8],covidDate[710,8],covidDate[740,8], covidDate[771,8],covidDate[801,8],covidDate[832,8],covidDate[863,8], covidDate[893,8],covidDate[924,8],covidDate[954,8],covidDate[985,8], covidDate[1016,8],covidDate[1044,8],covidDate[1075,8],covidDate[1105,8], covidDate[1136,8],covidDate[1166,8],covidDate[1197,8],covidDate[1228,8], covidDate[1258,8],covidDate[1289,8]) # Merge the above four data into the twocountries data frame twocountries<-data.frame(bymonth,chinatotalcases,chinatotaldeaths,uktotalcases,uktotaldeaths) twocountries # Because ggPlot requires three columns of data, twocountries is transformed # NB - For this part you need to install the reshape2 package twocountries<-melt(twocountries,id="bymonth",na.rm = TRUE) # Display twocountries twocountries # ggplot takes month as x and quantity as y and classifies according to variable ggplot(data = twocountries,aes(x=bymonth,y=value,group=variable,color=variable,shape=variable))+ geom_point()+ geom_line()+ geom_line()+ xlab("month")+ ylab("number")+ theme_bw()+ scale_x_continuous(limits = c(1,22),breaks = seq(1,22,1)) ### 3. Present descriptive statistics # Create labels for china four month new data frame fourmonth<-c(rep("ChinaApr2020",times=30),rep("ChinaOct2020",times=31), rep("ChinaApr2021",times=30),rep("ChinaOct2021",times=31)) # The total number of new cases in China over four months chinafourmonth<-c(covidDate[c(71:100),6],covidDate[c(254:284),6], covidDate[c(436:465),6],covidDate[c(619:649),6]) chinafourmonth<-data.frame(month=fourmonth,data=chinafourmonth) # Create labels for uk four month new data frame fourmonth<-c(rep("UKApr2020",times=30),rep("UKOct2020",times=31), rep("UKApr2021",times=30),rep("UKOct2021",times=31)) # The total number of new cases in uk over four months ukfourmonth<-c(covidDate[c(711:740),6],covidDate[c(894:924),6], covidDate[c(1076:1105),6],covidDate[c(1259:1289),6]) ukfourmonth<-data.frame(month=fourmonth,data=ukfourmonth) # Create data frame from labels chinafourmonth and ukfourmonth data fourmonthnew<-rbind(chinafourmonth,ukfourmonth) # Create labels for descStats data frame descStatsLabels<-c("Mean", "SD", "Median", "Max", "Min", "Range") # Four months of descriptive statistics were calculated for each country ChinaApr2020<-generateDescStats(fourmonthnew[1:30,2]) ChinaOct2020<-generateDescStats(fourmonthnew[31:61,2]) ChinaApr2021<-generateDescStats(fourmonthnew[62:91,2]) ChinaOct2021<-generateDescStats(fourmonthnew[92:122,2]) UKApr2020<-generateDescStats(fourmonthnew[123:152,2]) UKOct2020<-generateDescStats(fourmonthnew[153:183,2]) UKApr2021<-generateDescStats(fourmonthnew[184:213,2]) UKOct2021<-generateDescStats(fourmonthnew[214:244,2]) # Create data frame from labels and descriptive statistics data CovidSummary<-data.frame(Lable=descStatsLabels, ChinaApr2020,ChinaOct2020,ChinaApr2021,ChinaOct2021, UKApr2020,UKOct2020,UKApr2021,UKOct2021) CovidSummary ### 4. t-test ### # t-test: The number of new cases in a two-month period # Extract the number of new cases for each month october20<-covidDate[894:924,6] october21<-covidDate[1259:1289,6] # Is each data sample normal? Use Shapiro test to find out shapiro.test(october20) shapiro.test(october21) # Both of these show as p>0.05, so they are normally distributed # Perform t-test for independent samples # HO: There is no difference between the means, i.e. no significant difference in the number of new cases # H1: There is a difference between the means, i.e. significant differences in the number of new cases t.test(october20, october21) # The calculated p-value for this is 2.2e-16. This is smaller than 0.05, # so we cannot accept the null hypothesis at the p=0.05 level. # We therefore reject the null hypothesis that There is a significant difference # in the number of new cases in the UK between October 2020 and October 2021. ### 5. ANOVA TEST ### # Difference between April 2020, October 2020, April 2021 and October 2021 # H0: There is no difference between April 2020, October 2020, April 2021 and October 2021 # h1: There is a significant difference between at least two of the groups anovaData<-covidDate[c(71:100,254:284,436:465,619:649),6] chinamonth<-c(rep("April2020",times=30),rep("October2020",times=31), rep("April2021",times=30),rep("October2021",times=31)) anovaData<-data.frame(month=chinamonth,number=anovaData) # carry out ANOVA test # compute the analysis if variance resaov<-aov(anovaData$number ~ anovaData$month,data = anovaData) # summary of the analysis summary(resaov) # Df Sum Sq Mean Sq F value Pr(>F) # anovaData$month 3 17843 5948 5.202 0.00207 ** # Residuals 118 134921 1143 # --- # Signif. codes: 0 ??***?? 0.001 ??**?? 0.01 ??*?? 0.05 ??.?? 0.1 ?? ?? 1 # There is a significant difference between groups at the p=0.01 level # we can reject the null hypothesis and accept the alternative hypothesis # It is therefore necessary to perform post-hoc tests to find out where there difference # perform Tukey Pair-wise post-hoc analysis TukeyHSD(resaov) # Tukey multiple comparisons of means # 95% family-wise confidence level # Fit: aov(formula = anovaData$number ~ anovaData$month, data = anovaData) # $`anovaData$month` # diff lwr upr p adj # April2021-April2020 -29.966667 -52.719262 -7.214072 0.0045305 # October2020-April2020 -26.262366 -48.830726 -3.694005 0.0155795 # October2021-April2020 -10.068817 -32.637178 12.499543 0.6514888 # October2020-April2021 3.704301 -18.864059 26.272661 0.9736185 # October2021-April2021 19.897849 -2.670511 42.466210 0.1044000 # October2021-October2020 16.193548 -6.189061 38.576158 0.2398662 # The Tukey test shows that there are two significant difference at the p=0.05 level # between April2021-April2020 and October2020-April2020 # but not between other pairs of groups