Case-Study-Electricity.Rmd

---
title: "`data.table` Case Study - Electricity"
author: "Christos Gkenas"
date: "`r format(Sys.time(), '%a %d %b %Y (%H:%M:%S)')`"
output:
  html_document:
    highlight: tango
    theme: united
    toc: yes
    toc_depth: 3
    toc_float:
      collapsed: no
      smooth_scroll: no
  pdf_document:
    toc: yes
    toc_depth: '3'
---


```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, out.width="100%")

# Load all the libraries that you will use once here
library(here)
library(data.table)
library(countrycode)
library(ggplot2)
library(magrittr)
library(ggrepel)
library(knitr)
library(kableExtra)

# Changing default settings here.
theme_set(theme_classic(base_size = 16))
```

# Case Study: Electric power consumption

# Objectives

*Describe the data and its limitations*

## Questions

1)By continent, which countries have the lowest and highest consumption?
2) By continent, which countries have the lowest and highest percentage change in consumption since 2000?
3) How does Portugal compare to other EU Countries?



# Data Details

Download "Electric power consumption (kWh per capita)" data.
- World Bank Indicator
: 'EG.USE.ELEC.KH.PC'.
 
- https://data.worldbank.org/indicator/EG.USE.ELEC.KH.PC




# Data Processing

```{r}
list.files()
api <- fread(here("data", "Electricity_data.csv"),
             skip = 4, header = TRUE,
            check.names = TRUE)
api
View(api)
```


```{r readData, warning=FALSE}
# Add Continent to WiP dataset.
cl <- as.data.table(codelist)[, .(continent, wb)]

apicl <- merge(api, cl, by.x = "Country.Code", by.y = "wb", all.x = TRUE)
setnames(apicl, c("continent"), c("Continent"))
View(apicl)

ED <- melt(apicl,
           id.vars = c("Continent", "Country.Name", "Country.Code", "Indicator.Code"),
           measure = patterns("^X"),
           variable.name = "YearC",
           value.name = c("elcKW"),
           na.rm = TRUE)
View(ED)

# Note cLabel will be used to label the ends of lines in plots.
ED[, `:=` (Year = as.numeric(gsub("^X", "", YearC)))][
  , fYear := factor(Year)][
  , maxYear:=max(Year), by = .(Country.Name)][
    Year==maxYear, cLabel:=Country.Name][
      , c("maxYear", "YearC"):=NULL]

setcolorder(ED, c("Indicator.Code", "Continent", "Country.Name", "Country.Code", 
                  "Year"))

setkeyv(ED, c("Indicator.Code", "Continent", "Country.Name", "Country.Code", "Year"))

ED
View(ED)
```

# Analysis


## Portugal - Trends and Comparisons

> What are the time trends for Portugal?

Let's start by looking at a plot of how Portugal is performing over time.

```{r PTplot, warning=FALSE}
ED[Country.Name %in% "Portugal"] %>% 
ggplot(aes(Year, elcKW)) +
  geom_line(colour = "blue") + 
  geom_point(colour = "blue") +
  scale_x_continuous(breaks=seq(1960, 2015, 5)) +
  scale_y_continuous(limits=c(0, 5000)) +
  ggtitle("Portugal") +
  xlab("Year") +
  ylab("KW") 
```

**Interpretation**

In 1990 Portugal had 7.6% women in parliament (i.e. 12.2 men for each woman), which 
increased to 34.8% (i.e. 1.87 men for each woman) in 2018. This still falls short 
of 50% (i.e. point of gender parity in parliament).

Let's now plot how Portugal compares with other European countries. As a reference
the European Union and World averages. Finland, Hungary, Sweden and Romania are highlighted
for discussion.

```{r PTvsEurope, warning=FALSE}
ED[Continent == "Europe"] %>%
  ggplot(aes(Year, elcKW, group=Country.Name, colour = Country.Name, label = cLabel)) +
  geom_line(colour="grey90") +
  geom_line(data = ED[Country.Name %in% c("Portugal", "European Union", "World", 
                                          "Sweden", "Finland", "Hungary", "Romania")],
            aes(colour = Country.Name)) +
  scale_x_continuous(breaks=seq(1960, 2015, 5)) +
  scale_y_continuous(limits=c(0, 40000), breaks=seq(0, 40000, by= 5000)) +
  expand_limits(x = 2015) +
  geom_label_repel(data = ED[Country.Name %in% c("Portugal", "European Union", "World", 
                                                 "Sweden", "Finland", "Hungary", "Romania")],
                   xlim = c(2005, 2015)) +
  theme(legend.position = "none") +
  ggtitle("Portugal compared to European countries") +
  xlab("Year") +
  ylab("KW")
```

**Interpretation**

Portugal has had more women in parliament than the European Union average since 2007
and since around 2000 compared to the world average. Hungary and Romania both had 
a higher percentage of women in parliament in 1990 (around the end of the Cold War) 
than they have had since. The key point to note is that none of these countries 
reaches equality between males and females in parliament, although Sweden and 
Finland come closest.


## Highest Percentages 

> Which countries have the highest percentage of women in parliament by year?  How do continents compare?

```{r highestContinent, collapse=TRUE}
hgWiP <- WP[!is.na(Continent)][
  order(Continent, Year, -pctWiP), head(.SD, 1), by = .(Continent, Year)][
  , CountryWiP := sprintf("%3.1f%% - %s", pctWiP, Country.Name)][
  , .(Continent, Year, CountryWiP)]

hgCont <- dcast(hgWiP, Year ~ Continent, value.var = "CountryWiP") 

hgCont %>% 
  kable(align="clllll") %>% 
  kable_styling(bootstrap_options = "striped")
```

## No Women in Parliament

Which countries have no (0%) women in parliament?

```{r zeroWiP, collapse=TRUE}
zeroWiP <- WP[!is.na(Continent) & pctWiP==0]

zeroYear <- zeroWiP[order(Country.Name), .(Year, Country.Name)][
  , `:=`(N = max(.N)), .(Year)][
  , toString(paste0('"', Country.Name, '"')), by = .(Year, N)][order(Year)]
setnames(zeroYear, c("V1"), c("zeroWiP"))

zeroYear %>% 
  kable(align="lrl", col.names = c("Year", "Num", "Countries without any Women in Parliament")) %>% 
  kable_styling(bootstrap_options = "striped") %>% 
  column_spec(1, bold = T)
```


## Global Trends

> What are the global trends over time?

In the following each country has been plotted as a line and the world "average" is
highlighted in blue. 

```{r globalTrendsLines, warning=FALSE}
WP[!is.na(Continent)] %>%
  ggplot(aes(Year, pctWiP, group = Country.Name, label = cLabel)) +
  geom_line(colour="grey90") +
  geom_line(data=WP[Country.Name=="World"], colour="blue") +
  expand_limits(x = 2021) +
  geom_label_repel(data = WP[Country.Name=="World"], 
                   xlim = c(2018, 2021), colour = "Blue") +
  theme(legend.position = "none") +
  scale_x_continuous(breaks=seq(1990, 2021, 5)) +
  scale_y_continuous(limits=c(0, 70), breaks=seq(0, 70, by=10)) +
  ggtitle("Women in Parliament: Global Trends") +
  ylab("% Women in Parliament")
```

A box whisker plot is better suited to present the variation (uncertainty) over
time.

```{r globalTrendsBoxplot, message=FALSE}
WP[!is.na(Continent)] %>%
  ggplot(aes(fYear, pctWiP)) +
  geom_boxplot(width = 0.25, outlier.size = .5) +
  scale_x_discrete(breaks=seq(1990, 2020, 2)) + 
  scale_y_continuous(limits=c(0, 70), breaks=seq(0, 70, by=10)) +
  ggtitle("Box Whisker Plot for countries") +
  ylab("% Women in Parliament") +
  xlab("Year") 
```

### Continents

We can look at the global trends by continent for comparison.

```{r continent TrendsBoxplot, message=FALSE}
# First add observations for "World" as a continent.
wWP <- rbindlist(list(WP[!is.na(Continent)], 
                 WP[!is.na(Continent)][, Continent:="World"]))

wWP %>%
  ggplot(aes(fYear, pctWiP)) +
  geom_boxplot(width = 0.4, outlier.size = .5) +
  geom_line(data = WP[Country.Name=="World"][
    , .(fYear, pctWiP, Country.Name)], 
    aes(fYear, y = pctWiP, group = Country.Name), colour = "darkred") +
  scale_x_discrete(breaks=seq(1990, 2020, 5)) + 
  scale_y_continuous(limits=c(0, 70), breaks=seq(0, 70, by=20)) +
  facet_wrap(~Continent) +
  ggtitle("Box Whisker Plot for countries") +
  ylab("% Women in Parliament") +
  xlab("Year")+
  theme_classic(base_size=11) +
  labs(caption = "The red line is the world average.")
```


**Interpretation**

Although the world average is going up there is still large variation between countries. 
The interpretation by continent is similar. Note that in earlier years fewer countries 
provided data so there is likely to be some bais in the plots above.

# Conclusions

```{r worldConclusions, echo=FALSE}
pctWiPLast <- WP[Country.Name=="World"][order(Year), .SD[.N]][, pctWiP]
ratioWiPLast <- WP[Country.Name=="World"][order(Year), .SD[.N]][, Ratio]
yearLast <- WP[Country.Name=="World"][order(Year), .SD[.N]][, Year]
```

In `r yearLast`, globally there were `r sprintf("%3.1f%%", pctWiPLast)` women in parliament, 
i.e, for every woman in parliament there was `r sprintf("%2.1f", ratioWiPLast)` men 
in parliament. Whichever way we look at the data the conclusions is that, in general, 
the percentage of women in parliament is increasing but that gender parity in most 
parliaments is still far-off.

<hr>

# Session Information

```{r sessionInfo, echo=TRUE}
sessionInfo()
```