Unemploymentrate_jobagency_comparison

Unemploymentrate_jobagency_comparison

Name of QuantLet : Unemploymentrate_jobagency_comparison

Published in : Frühsignale für Änderungen von Konjunkturindikatoren durch Analysen von Big Data

Description : 'Shows how to get a time series from Google Trend with R. Shows some outputs of
linear regression with a business cycle time series (Unemployment rate). Main focus points on
visualization of the two time series.'

Keywords : 'regression, linear-regression, time-series, scatterplot, boxplot, plot, graphical
representation, data visualization, financial'

Author : Daniel Jacob

Datafile : Arbeitslosenquote.csv

Example: 
- 1: Plot of the German unemployment rate.
- 2: Boxplot of the German unemployment rate.
- 3: Scatterplot of the official German unemployment rate.
- 4: Graphical representation of the linear-regression (Residuals).
- 5: Graphical representation of the linear-regression (Normal Q-Q).
- 6: Biplot of ''Job centre'' (Arbeitsamt) and unemployment rate.
- 7: Summary of the linear-regression Output.

R Code:

# clear variables and close windows
rm(list = ls(all = TRUE))
graphics.off()

# install and load packages
libraries = c("plotrix", "devtools")
lapply(libraries, function(x) if (!(x %in% installed.packages())) {
    install.packages(x)
})
lapply(libraries, library, quietly = TRUE, character.only = TRUE)

# devtools needed
install_github("googletrend", "okugami79")

library(googletrend)

# See the Message in the 'Console' --> 'your webbrowser download directory path :
# 'path' is your working directory for Google Trend
googletrend::setdownloaddir("path")

# Load Unemployment rate (quarterly adjusted)
Arbeit = read.csv("Arbeitslosenquote.csv", header = TRUE, sep = ";", dec = ",")

# Plot 1
plot(Arbeit[, "Arbeitslosenquote"], type = "l", pch = 1, col = "red", xaxt = "n", 
    xlab = "Date", ylab = "unemployment rate", ylim = c(5, 13), lwd = 2, main = "Plot of the German unemployment rate")
# 2. X-Axis = Date
axis(1, at = c(1:nrow(Arbeit)), labels = Arbeit[, "Jahre"])

# Plot 2 Descriptive Statistics
summary(Arbeit)
dev.new()
boxplot(Arbeit[, -1], ylab = "Unemployment rate")

# New Data with Word 'labor bureau (Arbeitsamt)' using Google Trend
arbeitsamt = gettrend(keyword = "Arbeitsamt", geo = "DE", use.monthly = TRUE)

# Divide Numbers by Ten (for a better overview)
ten = matrix(10, ncol = 1, nrow = 142)
arbeitsamt2 = cbind(arbeitsamt, ten)
arbeit_ten = apply(arbeitsamt2[, -1], 2, function(x) (x/ten))
arbeit_ten = as.data.frame(arbeit_ten)
arbeitsamt3 = cbind(arbeitsamt[, 1], arbeit_ten[, 1])
arbeitsamt4 = cbind(arbeitsamt2, arbeitsamt3[, 2])
arbeitsamt5 = arbeitsamt4[, -c(2, 3)]

# Linear Regression (OLS-Method) With Unemployment rate vs. Google Words
lm_unemp_arbeitsamt = lm(Arbeit[, 2] ~ arbeitsamt5[13:132, 2])
summary(lm_unemp_arbeitsamt)

# Plot 3
dev.new()
plot(Arbeit[, 2] ~ arbeitsamt5[13:132, 2], ylab = "official Unemployment rate", xlab = "Frequency of the term 'job centre'", 
    main = "Scatter plot")
abline(lm_unemp_arbeitsamt, col = "red")

# Plot 4: more graphical representations of the Linear Regression
dev.new()
plot(lm_unemp_arbeitsamt)

# Plot 5: 'job centre' (Arbeitsamt) vs. Unemployment rate
dev.new()
twoord.plot(lx = arbeitsamt5[13:132, 1], ly = Arbeit[, 2], rx = arbeitsamt5[13:132, 
    1], ry = arbeitsamt5[13:132, 2], xtickpos = as.numeric(arbeitsamt5[13:132, 1]), 
    xticklab = as.character(arbeitsamt5[13:132, 1]), xlab = "Date", ylab = "Unemployment rate", 
    rylab = "frequency of arbeitsamt", main = "Unemployment rate vs. Frequency of the term 'job centre'")
legend("topright", col = c("black", "red"), pch = 1:2, c("Unemployment rate", "job centre"), 
    lty = 1, lwd = 1, cex = 1, bty = "n")