Machine Learning Made Easy

Monday, May 25, 2020

Blog 26: Time Series Plots using ggplot

Time Series Plots

Introduction

The only analysis where plots drive the specific model components is time series. Just by looking at the plots, one can comment about the seasonality ,trend and get an idea of how to create a model around the data. In this blog we will look at how to plot a time series variable using ggplot.The dataset we are going to take for our analysis is-Winning times (in minutes) for the Boston Marathon Men’s Open Division. 1897-2016. which is available from fpp2 package

Installing the library: dplyr,tidyr and Ecdat package

package.name<-c("dplyr","tidyr","fpp2","ggplot2")

for(i in package.name){

  if(!require(i,character.only = T)){

    install.packages(i)
  }
  library(i,character.only = T)

}


# fpp2 package has the 'marathon' data
data(marathon)
df<-marathon
head(df)

Time Series:
Start = 1897 
End = 1902 
Frequency = 1 
[1] 175.1667 162.0000 174.6333 159.7333 149.3833 163.2000

Lets create the data frame and plot the time series attribute

interim.df<-data.frame(TimeUnits=seq(start(df)[1],end(df)[1],by=frequency(df)),
                       Values=df)

ggplot(interim.df, aes(x = TimeUnits, y = Values)) + 
  geom_line(aes(color = "orange"), size = 1) +
  scale_color_manual(values = c("orange")) +
  theme_minimal()

Look how we created a data frame from the time series attribute by using start,end and frequency property of time series data

Area Plot

Lets also try and plot the time series variable using area plots. This is also used to represent long standing view of stock prices,inventory levels, asset value,etc

ggplot(interim.df, aes(x = TimeUnits, y = Values)) + 
  geom_area(aes(color = "orange",fill="orange"), 
            alpha = 0.5, position = position_dodge(0.8))+
  scale_color_manual(values = c("orange")) +
  scale_fill_manual(values = c("orange"))