Time Series Data Analysis in R

automation Importing multiple sheets

What if you had multiple sheets in the same excel file, and you needed to import all of them and apply some sort of descriptive statistics.

We automate this process by using a bit of programming. There are quite a few things that we can use to achieve this task . Three of them are:

1. Using loops
2. Using lapply
3. Using the RIO package

The RIO package

Plot multiple data.frames

main.r

ggplot(data_keys, aes(x = Date)) +
	geom_line(aes(y=Min, colour="Min"),linetype="dashed") +
	geom_line(aes(y=Max, colour="Max"),linetype="dashed") +
	geom_line(aes(y=Modal, colour="Modal"),linetype="dashed") +
	scale_color_manual(name = 'Price Series', values=c("Min"="tomato", "Max"="royalblue", "Modal"="black")) + 
	facet_wrap(vars(keys))

Separate Plots

separate.r

for(i in names(data)) {
df1 <- as.data.frame(data[[i]])
plotdf1 <- ggplot(df1, aes(x = Date)) +
	geom_line(aes(y=Min, colour="Min"),linetype="dashed") +
	geom_line(aes(y=Max, colour="Max"),linetype="dashed") +
	geom_line(aes(y=Modal, colour="Modal"),linetype="dashed") +
	scale_color_manual(name = 'Price Series', values=c("Min"="tomato", "Max"="royalblue", "Modal"="black")) + 
	labs(x="Date",y="Value",title=paste(i,"Prices")) +
	theme(legend.position="inside",legend.position.inside=c(0.09,0.828))
 
png(file.path(img,paste(i,'png',sep='.')))
print(plotdf1)
dev.off()
}

def.off() is used to signify that there has been an end of input into the png file, more like a flush function that is used in file handling.

Correlation plots

combine.r

df <- as.data.frame(do.call(cbind,data))
names(df) <- sub(" .*", "", names(df))
summary(df)
stargazer(data.frame(df),type="text", summary=TRUE)
 
df_nodate = df[, -c(1,5,9,13)]
(cor_df_nodate <- round(cor(df_nodate),2))
corrplot(cor_df_nodate, type="upper", order="h)

cbind does a sort of column bind, so that information is placed in a single column¹ df_nodate because correlation works only with numerical values, we will have to remove any dates… 1, 2.3.4, 5 , 6.7.8, 9 , 10.11.12, 13, 14.15.16

corrplot is used to create a correlation plot.

Footnotes

1. think of flex-col, there is also rbind for rows ↩