## plot a geom_col and a geom_bar to see the differences
ggp_bar <- ggplot(data = gapminder, aes(x=continent)) +
    geom_bar()

## what is 'count' in this graph?
ggp_bar

## geom_col
ggp_col <- ggplot(data = gapminder %>%
                       filter(continent == "Oceania", year == 2007),
                  aes(x=country, y=pop)) + # can do dataframe manipulation within the data call with pipes.
    geom_col()
## here, y axis is the value for the variable you pass in the aesthetic.
ggp_col

## Now add in more of the layers to one of the plots above

ggp_col2 <- ggplot(data = gapminder %>%
                       filter(continent == "Oceania", year == 2007), aes(x=country, y=pop)) +
    geom_col(aes(fill = country)) + # notice it adds in a legend automatically
    scale_y_continuous(breaks = seq(0,21000000, 2000000)) + #try changing the breaks
    ggtitle("Population of Oceania by country in 2007") +
    labs(y = "Population", x = "Country", fill = "Country") +
    theme_bw()

ggp_col2

## plot a scatter plot with geom_point
## work through this a bit at a time to understand what each section does

ggp_scatter <- ggplot(data = gapminder, # data
              aes(x = gdpPercap, y = lifeExp, colour = continent)) + #aesthetics
    geom_point(alpha = 0.5) + #geometries
    geom_smooth(method = NULL, se = T) +
    facet_wrap(~continent, scales = "free") + #facets
    labs(title = "Life expectancy at birth vs. Per capita GDP", x = "gdpPercap", y = "lifeExp", colour = "Continent") + #labels
    theme_light() + #themes
    theme(axis.text.x = element_text(angle = 45),
          axis.text = element_text(size = 7),
          legend.position = "bottom") +
    NULL #good to include a NULL because of the + convention.

ggp_scatter

## repeat these plots using different variables

## plot a histogram showing distribution of life expectancy
ggp_hist <- ggplot(data = gapminder, aes(x = lifeExp)) +
    geom_histogram(aes(y = ..density..), binwidth = 2, colour = "black", fill = "lightblue") + # change binwidth or substitute out for bins =
    geom_density(aes(colour = continent, fill = continent), alpha = 0.1) +
    labs(title = "Life Expectancy", x = "Life Expectancy", y = "Density") +
    theme_light() +
    NULL
## instead of having different density plots overlaying the same graph, try facet_wrap().
ggp_hist

# ggp_hist2 <- ggplot() +
#     geom_histogram() +
#     geom_density() +
#     NULL

#ggp_hist2
## make another histogram with different variables

# ggp_hist3 <- ggplot() +
#     geom_histogram() +
#     NULL

#ggp_hist3

## plot a boxplot

ggp_box <- ggplot(data = gapminder, aes(x = fct_reorder(continent, lifeExp), y = lifeExp, fill = continent)) + # useful reordering function suppled to the x aesthetic. Remove it and see the effects.
    geom_boxplot(outlier.color = NA) + #removed outliers in this plot because we are layering on geom_point() of the data, try commenting out the geom_point() and removing the outlier.color = NA command to see how outliers are presented by default.
    geom_jitter(alpha = 0.2) + #geom_jitter is a variation on geom_point(). Swap out for geom_point() to see the effects.
    scale_y_continuous(breaks = round(seq(min(gapminder$lifeExp), max(gapminder$lifeExp), by = 10),0)) + # seq() is for sequence, change the by = 10 to a different value to see the effects.
    coord_flip() + # comment coord_flip() out to see the effects.
    labs(title = "Boxplot of life expectancy by continent", x = "Life Expectancy", y = "Continent") +
    theme_light() +
    theme(legend.position = "none") + # removed the legend as not visually necessary in this plot.
    NULL

ggp_box

## wrangle gapminder data to find top 5 countries in the year 2007 with the highest life expectancy and create a new boxplot.
top5 <- gapminder %>%
    filter() %>% #filter for year
    arrange() %>% #arrange by life expectancy
    head(5)

## wrangle some features
## calculate rate change from previous year per country
df_rate_change <- gapminder %>%
    janitor::clean_names() %>% # good package and function to make column names more coding friendly.
    group_by(country) %>%
    arrange(year) %>%
    mutate(pc_lifeexp_change = round(((life_exp/ lag(life_exp)) -1) * 100, 1)) %>% # here, rounding is done to 1 decimal place.
    mutate(pc_pop_change = round(((pop / lag(pop)) -1) * 100, 1)) %>%
    mutate(pc_gdp_cap_change = round(((gdp_percap / lag(gdp_percap)) -1) * 100, 1)) %>% #changes here are * 100 to make %
    arrange(country)

## add in any other additional features you would like to visualise with mutate()
head(df_rate_change)

## plot a line graph showing the changes in life expectancy per year for 5 countries
## find the 5 countries with the lowest population in 2007 per continent
low_pop_2007 <- df_rate_change %>%
    filter(year == 2007) %>%
    group_by(continent) %>%
    arrange(life_exp) %>%
    slice(1) %>% # slice() takes the first row per group - here it is important that the previous steps are correct. head() will only take the 1st row of the dataframe.
    ungroup() %>%
    pull(country) #pull() will return a list of the countries, use class(low_pop_2007) to see the structure of the outcome.

high_pop_2007 <- df_rate_change %>%
    filter(year == 2007) %>%
    group_by(continent) %>%
    arrange(desc(life_exp)) %>% # nesting desc() within arrange() will arrange in descending order.
    slice(1) %>%
    ungroup() %>%
    pull(country)

countries_to_plot <- c(as.character(low_pop_2007), as.character(high_pop_2007)) # have to change to character type from factors. See what happens if you remove the as.character() calls.
countries_to_plot

ggp_pc_rate <- ggplot(data = df_rate_change %>%
                  filter(country %in% countries_to_plot), aes(x = year, colour = country)) +
    geom_point(aes(y = pc_gdp_cap_change)) +
    geom_line(aes(y = pc_gdp_cap_change), linetype = "dashed") + # difficult to add in the legend of linetypes when layering graphs like this.
    geom_point(aes(y = pc_pop_change)) +
    geom_line(aes(y = pc_pop_change)) +
    facet_grid(rows = vars(continent)) +
    scale_x_continuous(breaks = seq(min(df_rate_change$year), max(df_rate_change$year), by = 5)) +
    scale_y_continuous(breaks = seq(-50, 50, by = 25)) +
    coord_cartesian(xlim = c(1957, 2007), y = c(-50,50)) +
    labs(title = "Change in GDP per capita and population time", x = "Year", y = "Percent change (%)") +
    theme_light() +
    NULL

ggp_pc_rate

## ggplot loves long data. Let's wrangle the data to be in long format.

df_rate_change_pivot <- df_rate_change %>%
    filter(country %in% countries_to_plot) %>%
    select(country,
           continent,
           year,
           pc_pop_change,
           pc_gdp_cap_change) %>%
    pivot_longer(c(pc_pop_change, pc_gdp_cap_change),
                   names_to = "pc_change_type",
                   values_to = "pc_change")

## now make the graph
ggp_pc_rate_pivot <- ggplot(data = df_rate_change_pivot, aes(x = year, y = pc_change, colour = country)) +
    geom_point() +
    geom_line(aes(linetype = pc_change_type)) +
    facet_grid(rows = vars(continent)) +
    scale_x_continuous(breaks = seq(min(df_rate_change$year), max(df_rate_change$year), by = 5)) +
    scale_y_continuous(breaks = seq(-50, 50, by = 25)) +
    coord_cartesian(xlim = c(1957, 2007), y = c(-50,50)) +
    labs(title = "Change in GDP per capita and population time", x = "Year", y = "Percent change (%)", colour = "Country", linetype = "% Change Variable") +
    theme_light() +
    NULL

ggp_pc_rate_pivot

ggp_for_plotly <- ggplot(data = df_rate_change %>%
                  filter(country %in% countries_to_plot), aes(x = year, y = life_exp, colour = country)) +
    geom_point(aes(text = paste0("Country: ",country,
                                 "<br>Year: ",year,
                                 "<br>Life Expectancy: ",life_exp,
                                 "<br>Life Expectancy (% change): ", pc_lifeexp_change))) +
    geom_line() +
    scale_x_continuous(breaks = seq(min(df_rate_change$year), max(df_rate_change$year), by = 5)) +
    labs(title = "Change in life expectancy over time", x = "Year", y = "Life expectancy (years)", colour = "Country") +
    theme_light() +
    NULL


interactive_graph <- plotly::ggplotly(ggp_for_plotly, tooltip = "text")
interactive_graph