covid19projectv6.Rmd

---
title: "Covid-19 Project"
author: "Deeksha Daga"
date: "May 5, 2020"
output:
  prettydoc::html_pretty:
    theme: hpstr
    highlight: github
vignette: >
  %\VignetteIndexEntry{Creating Pretty Documents from R Markdown - The HPSTR Theme}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

```{r echo=FALSE }
library(gtools)
library(rjson)
library(plyr)
library(Metrics)
my.JSON <- fromJSON(file="https://pomber.github.io/covid19/timeseries.json")
India_json <- lapply(my.JSON$India, function(x){ unlist(x)})
India_df <- rbind.fill(lapply(India_json,function(x) do.call("data.frame", as.list(x))))
India_df <- do.call("rbind", India_json)
India_df <- as.data.frame(India_df)

cdate <- as.Date(India_df$date,format = "%Y-%m-%d")
cconfirmed <- as.numeric(India_df$confirmed)
cdeaths <- as.numeric(India_df$deaths)
crecovered <- as.numeric(India_df$recovered)

world_json <- lapply(my.JSON, function(x){unlist(x)})
world_df <- rbind.fill(lapply(world_json,function(x) do.call("data.frame",as.list(x))))
world_df <- do.call("rbind", world_json)
world_df <- as.data.frame(world_df)

wdate <- as.Date(world_df$date,format = "%Y-%m-%d")
wconfirmed <- as.numeric(world_df$confirmed)
wdeaths <- as.numeric(world_df$deaths)
wrecovered <- as.numeric(world_df$recovered)
#No. of days since 22-jan-2020
current = as.Date(Sys.Date())
start = as.Date("2020-01-22")
today = as.numeric(current - start)
options(scipen=999)
India_df[59,3] <- as.numeric(4)
```

```{r echo=FALSE}
India_df1 <- cbind(India_df,newconfirmed=0,newdeaths=0,newrecovered=0)
India_df1$date <- as.Date(India_df1$date,format = "%Y-%m-%d")
dcdate <- as.Date(India_df1$date,format = "%Y-%m-%d")
India_df1$confirmed<- as.numeric(India_df1$confirmed)
India_df1$deaths <- as.numeric(India_df1$deaths)
India_df1$recovered <- as.numeric(India_df1$recovered)
i=2
while (i != today+1){
  India_df1[i,5] <-India_df1[i,2]-India_df1[(i-1),2]
  India_df1[i,6] <-India_df1[i,3]-India_df1[(i-1),3]
  India_df1[i,7] <-India_df1[i,4]-India_df1[(i-1),4]
  i=i+1
}
newconfirmed<- as.numeric(India_df1$newconfirmed)
```

### Graph showing total confirmed cases in India:

```{r echo=FALSE}
# polynomial regression
cfit = lm(cconfirmed~poly(cdate,6))

#scatter plot
par(las=2,mgp=c(5,1,0),mar=c(6,6,6,6))
plot(cdate,cconfirmed,pch = 16,xlab = "date", col = "blue",xaxt = "n"
     ,ylab = "no. of cases ", main = "confirmed cases in india")
points(cdate,fitted(cfit),col="red",pch=1)
axis.Date(1, at=seq(min(cdate),max(cdate),by=10), format="%d %b %y")
legend("topleft",c("Actual","Predicted"), 
       col=c("blue","red"), lwd=2)
```

```{r echo=FALSE}
#prediction
a = today+1; b = today+7
new <- data.frame(cdate = as.Date(a:b ,origin = "2020-01-21"))
pred_confirmed <-predict.lm(cfit,new,interval = "confidence",level = 0.95)
newindia_df <- cbind(new, pred_confirmed)
names(newindia_df)[3] <- "minimum"
names(newindia_df)[4] <- "maximum"
names(newindia_df)[1] <- "date"
newindia_df$fit <- round(newindia_df$fit)
newindia_df$minimum <- round(newindia_df$minimum)
newindia_df$maximum <- round(newindia_df$maximum)
```
### Predictions for total confirmed cases in India for next week:

```{r echo=FALSE}
knitr::kable(newindia_df,align = "crrr",format.args = list(big.mark = ",", 
  scientific = FALSE))
```

```{r echo=FALSE}
# polynomial regression
dcfit = lm(newconfirmed~poly(dcdate,6))

#scatter plot
par(las=2,mgp=c(5,1,0),mar=c(6,6,6,6))
plot(India_df1$date,India_df1$newconfirmed,pch = 16,xlab = "date", col = "blue",xaxt = "n"
     ,ylab = "no. of cases ", main = "new confirmed cases in india everyday")
points(India_df1$date,fitted(dcfit),col="red",pch=16)
axis.Date(1, at=seq(min(India_df1$date),max(India_df1$date),by=10), format="%d %b %y")
legend("topleft",c("Actual","Predicted"), 
       col=c("blue","red"), lwd=2)
```

```{r echo=FALSE}
#prediction
a = today+1; b = today+7
new1 <- data.frame(dcdate = as.Date(a:b ,origin = "2020-01-21"))
pred_confirmed <-predict.lm(dcfit,new1,interval = "confidence",level = 0.95)
newindia_df1 <- cbind(new,pred_confirmed)
names(newindia_df1)[3] <- "minimum"
names(newindia_df1)[4] <- "maximum"
names(newindia_df1)[1] <- "date"
newindia_df1$fit <- round(newindia_df1$fit)
newindia_df1$minimum <- round(newindia_df1$minimum)
newindia_df1$maximum <- round(newindia_df1$maximum)
```

### Predictions for new confirmed cases per day in India for next week:

```{r echo=FALSE}
knitr::kable(newindia_df1,align = "crrr",format.args = list(big.mark = ",", 
  scientific = FALSE))
```

```{r echo=FALSE}

errdate <- data.frame(date = as.Date(0:(today-101),origin = "2020-05-01"))
err <- cbind(errdate,errfit=0)
i=101
while (i != today+1){
  pdate <- cdate[c(i)]
  newdata <- subset(India_df, cdate<pdate)
  newdata$date <- as.Date(newdata$date,format = "%Y-%m-%d")
  ndate <- as.Date(newdata$date,format = "%Y-%m-%d")
  newdata$confirmed <- as.numeric(newdata$confirmed)
  newdata$deaths <- as.numeric(newdata$deaths)
  newdata$recovered <- as.numeric(newdata$recovered)
  fit = lm(newdata$confirmed~poly(ndate,6))
  new <- data.frame(ndate = as.Date(pdate ,origin = "2020-01-21"))
  pred_confirmed <-predict(fit,new)
  err[(i-100),2] <- pred_confirmed
  i=i+1
}
err$errfit <- round(err$errfit)
actual <- cconfirmed[101:today]
err <- cbind(err,actual)
err <- cbind(err,error = ape(actual,err$errfit)*100)
names(err)[2] <- "fit"
err$error <- round(err$error,2)
```

### Graph showing error percentage of our predictions for India:

```{r echo=FALSE}
par(las=2,mgp=c(5,1,0),mar=c(6,6,6,6))
plot(err$date,err$error,type ='l',xlab = "date",xaxt = "n", col = "blue"
     ,ylab = " error percentage ", main = "error percentage for india")
axis.Date(1, at=seq(min(err$date),max(err$date),by=1), format="%d %b %y")
```


### Error Analysis:

```{r echo=FALSE}
err$error <- paste(err$error,"%")
knitr::kable(err,align = "crrr",format.args = list(big.mark = ",", 
  scientific = FALSE))
```

```{r echo=FALSE}

errdate <- data.frame(date = as.Date(0:(today-101),origin = "2020-05-01"))
dcerr <- cbind(errdate,errfit=0)
i=101
while (i != today+1){
  pdate <- dcdate[c(i)]
  newdata <- subset(India_df1, dcdate<pdate)
  newdata$date <- as.Date(newdata$date,format = "%Y-%m-%d")
  ndate <- as.Date(newdata$date,format = "%Y-%m-%d")
  newdata$confirmed <- as.numeric(newdata$newconfirmed)
  newdata$deaths <- as.numeric(newdata$newdeaths)
  newdata$recovered <- as.numeric(newdata$newrecovered)
  dfit = lm(newdata$newconfirmed~poly(ndate,6))
  new <- data.frame(ndate = as.Date(pdate ,origin = "2020-01-21"))
  pred_confirmed <-predict(dfit,new)
  dcerr[(i-100),2] <- pred_confirmed
  i=i+1
}
dcerr$errfit <- round(dcerr$errfit)
actual <- India_df1$newconfirmed[101:today]
dcerr <- cbind(dcerr,actual)
dcerr <- cbind(dcerr,error = ape(actual,dcerr$errfit)*100)
names(dcerr)[2] <- "fit"
dcerr$error <- round(dcerr$error,2)
```

### Graph showing error percentage of our predictions of new cases everyday in India:

```{r echo=FALSE}
par(las=2,mgp=c(5,1,0),mar=c(6,6,6,6))
plot(dcerr$date,dcerr$error,type ='l',xlab = "date",xaxt = "n", col = "blue"
     ,ylab = " error percentage ", main = "error percentage for new cases in india per day")
axis.Date(1, at=seq(min(dcerr$date),max(dcerr$date),by=1), format="%d %b %y")
```

### Error Analysis:

```{r echo=FALSE}
dcerr$error <- paste(dcerr$error,"%")
knitr::kable(dcerr,align = "crrr",format.args = list(big.mark = ",", 
  scientific = FALSE))
```

```{r echo=FALSE}
t_case=c();t_dead=c();t_recovered=c();t_date=c(1:today)
count=0
cols = seq(from=2, length=today,by=4)
for (j in cols){
  total_cases = 0
  total_deaths = 0
  total_recovered = 0
  count = count + 1
  for(i in world_df[,j]){
    total_cases = total_cases + as.numeric(i)
  }
  for(i in world_df[,j+1]){
    total_deaths = total_deaths + as.numeric(i)
  }
  for(i in world_df[,j+2]){
    total_recovered = total_recovered + as.numeric(i)
  }
  t_case[count] = as.numeric(total_cases)
  t_dead[count] = as.numeric(total_deaths)
  t_recovered[count] = as.numeric(total_recovered)
}
wdate <- as.Date(t_date,origin = "2020-01-21")
wdf = cbind(t_date,t_case,t_dead,t_recovered)

wdf <- as.data.frame(wdf)
wconfirmed <- as.numeric(wdf$t_case)
wdeaths <- as.numeric(wdf$t_dead)
wrecovered <- as.numeric(wdf$t_recovered)
names(wdf)[1] <- "date"
names(wdf)[2] <- "confirmed"
names(wdf)[3] <- "deaths"
names(wdf)[4] <- "recovered"
```

### Graph showing total confirmed cases in World:

```{r echo=FALSE}
#polynomial regression
wfit = lm(wconfirmed~poly(wdate,6))

#scatter plot
par(las=2,mgp=c(5,1,0),mar=c(6,6,6,6))
plot(wdate,wconfirmed,pch = 16,xlab = "date", col = "blue",xaxt="n"
     ,ylab = "no. of cases", main = "confirmed cases in world")
points(wdate,fitted(wfit),col="red",pch=1)
axis.Date(1, at=seq(min(wdate),max(wdate),by=10), format="%d %b %y")
legend("topleft",c("Actual","Predicted"), col=c("blue","red"), lwd=2)
```

```{r pressure, echo=FALSE}
a = today+1; b = today+7
predict_date <- as.Date(a:b,origin="2020-01-20")
new <- data.frame(wdate = as.Date(a:b ,origin = "2020-01-22"))
preddate <- data.frame(wdate = as.Date(a:b ,origin = "2020-01-21"))
pred_wconfirmed <-predict.lm(wfit,new,interval = "confidence",level = 0.95)
newworld_df <- cbind(preddate, pred_wconfirmed)
names(newworld_df)[3] <- "minimum"
names(newworld_df)[4] <- "maximum"
names(newworld_df)[1] <- "date"
```

### Predictions for total confirmed cases in World for next week:

```{r echo=FALSE}
knitr::kable(newworld_df,align = "crrr",format.args = list(big.mark = ",", 
  scientific = FALSE))
```

```{r echo=FALSE}
World_df1 <- cbind(wdf,newconfirmed=0,newdeaths=0,newrecovered=0)
World_df1$date <- as.Date(World_df1$date,format = "%Y-%m-%d",origin = "2020-01-22")
dwdate <- as.Date(World_df1$date,format = "%Y-%m-%d")
World_df1$confirmed<- as.numeric(World_df1$confirmed)
World_df1$deaths <- as.numeric(World_df1$deaths)
World_df1$recovered <- as.numeric(World_df1$recovered)
i=2
while (i != today+1){
  World_df1[i,5] <-World_df1[i,2]-World_df1[(i-1),2]
  World_df1[i,6] <-World_df1[i,3]-World_df1[(i-1),3]
  World_df1[i,7] <-World_df1[i,4]-World_df1[(i-1),4]
  i=i+1
}
newwconfirmed<- as.numeric(World_df1$newconfirmed)
```

```{r echo=FALSE}
# polynomial regression
dwfit = lm(newwconfirmed~poly(dwdate,6))

#scatter plot
par(las=2,mgp=c(5,1,0),mar=c(6,6,6,6))
plot(World_df1$date,World_df1$newconfirmed,pch = 16,xlab = "date", col = "blue",xaxt = "n"
     ,ylab = "no. of cases ", main = "confirmed cases in World everyday")
points(World_df1$date,fitted(dwfit),col="red",pch=16)
axis.Date(1, at=seq(min(World_df1$date),max(World_df1$date),by=10), format="%d %b %y")
legend("topleft",c("Actual","Predicted"), 
       col=c("blue","red"), lwd=2)
```

```{r echo=FALSE}
#prediction
a = today+1; b = today+7
new1 <- data.frame(dwdate = as.Date(a:b ,origin = "2020-01-21"))
pred_confirmed <-predict.lm(dwfit,new1,interval = "confidence",level = 0.95)
newworld_df1 <- cbind(new1,pred_confirmed)
names(newworld_df1)[3] <- "minimum"
names(newworld_df1)[4] <- "maximum"
names(newworld_df1)[1] <- "date"
newworld_df1$fit <- round(newworld_df1$fit)
newworld_df1$minimum <- round(newworld_df1$minimum)
newworld_df1$maximum <- round(newworld_df1$maximum)
```

### Predictions for new confirmed cases per day in World for next week:

```{r echo=FALSE}
knitr::kable(newworld_df1,align = "crrr",format.args = list(big.mark = ",", 
  scientific = FALSE))
```
```{r echo=FALSE}
errdate <- data.frame(date = as.Date(0:(today-101),origin = "2020-05-01"))
err1 <- cbind(errdate,errfit=0)
i=101
while (i != today+1){
  pdate <- wdate[c(i)]
  newdata <- subset(wdf, wdate<pdate)
  newdata$date <- as.Date(newdata$date,format = "%Y-%m-%d",origin="2020-01-22")
  ndate <- as.Date(newdata$date,format = "%Y-%m-%d")
  newdata$confirmed <- as.numeric(newdata$confirmed)
  newdata$deaths <- as.numeric(newdata$deaths)
  newdata$recovered <- as.numeric(newdata$recovered)
  fit = lm(newdata$confirmed~poly(ndate,6))
  new <- data.frame(ndate = as.Date(pdate ,origin = "2020-01-21"))
  pred_confirmed <-predict(fit,new)
  err1[(i-100),2] <- pred_confirmed
  i=i+1
}
err1$errfit <- round(err1$errfit)
actual <- wconfirmed[101:today]
err1 <- cbind(err1,actual)
err1 <- cbind(err1,error = ape(err1$actual,err1$errfit)*100)
names(err1)[2] <- "fit"
err1$error <- round(err1$error,2)
```

### Graph showing error percentage of our predictions for the World:

```{r echo=FALSE}
par(las=2,mgp=c(5,1,0),mar=c(6,6,6,6))
plot(err1$date,err1$error,type ='l',xlab = "date",xaxt = "n" , col = "blue"
     ,ylab = " error percentage ", main = "error percentage for world")
axis.Date(1, at=seq(min(err1$date),max(err1$date),by=1), format="%d %b %y")
```

### Error Analysis:

```{r echo=FALSE}
err1$error <- paste(err1$error,"%")
knitr::kable(err1,align = "crrr",format.args = list(big.mark = ",", 
  scientific = FALSE))
```

```{r echo=FALSE}

errdate <- data.frame(date = as.Date(0:(today-101),origin = "2020-05-01"))
dwerr <- cbind(errdate,errfit=0)
i=101
while (i != today+1){
  pdate <- dwdate[c(i)]
  newdata <- subset(World_df1, dwdate<pdate)
  newdata$date <- as.Date(newdata$date,format = "%Y-%m-%d")
  ndate <- as.Date(newdata$date,format = "%Y-%m-%d")
  newdata$confirmed <- as.numeric(newdata$newconfirmed)
  newdata$deaths <- as.numeric(newdata$newdeaths)
  newdata$recovered <- as.numeric(newdata$newrecovered)
  dfit = lm(newdata$newconfirmed~poly(ndate,6))
  new <- data.frame(ndate = as.Date(pdate ,origin = "2020-01-21"))
  pred_confirmed <-predict(dfit,new)
  dwerr[(i-100),2] <- pred_confirmed
  i=i+1
}
dwerr$errfit <- round(dwerr$errfit)
actual <- World_df1$newconfirmed[101:today]
dwerr <- cbind(dwerr,actual)
dwerr <- cbind(dwerr,error = ape(actual,dwerr$errfit)*100)
names(dwerr)[2] <- "fit"
dwerr$error <- round(dwerr$error,2)
```

### Graph showing error percentage of our predictions for new cases in the World per day:

```{r echo=FALSE}
par(las=2,mgp=c(5,1,0),mar=c(6,6,6,6))
plot(dwerr$date,dwerr$error,type ='l',xlab = "date",xaxt = "n" , col = "blue"
     ,ylab = " error percentage ", main = "error percentage for world per day")
axis.Date(1, at=seq(min(dwerr$date),max(dwerr$date),by=1), format="%d %b %y")
```

### Error Analysis:

```{r echo=FALSE}
dwerr$error <- paste(dwerr$error,"%")
knitr::kable(dwerr,align = "crrr",format.args = list(big.mark = ",", 
  scientific = FALSE))
```