V42_ggplotly.Rmd

```{r include=FALSE}
knitr::opts_chunk$set(cache = T, warning = F, message = F, 
                      class.output = "output", out.width='100%', 
                      fig.asp = 0.5, fig.align = 'center')
```

## ggplotly

[Scatter plots with ggplot2 (plotly.com)](https://plotly.com/ggplot2/line-and-scatter/)

```{r include=FALSE}
library(tidyverse)
options(scipen = 999)
th <- theme_minimal() + 
  theme(
    title = element_text(family="Noto Sans CJK TC"), #all title elements
    text = element_text(family="Noto Sans CJK TC"),  #all text elements
    plot.title = element_text(face="bold"),
    axis.title.x = element_text(hjust=1, face="bold.italic"),
    axis.title.y = element_text(hjust=1, face="bold.italic"),
        )

```

### LINE CHART

[Line plots with R (plotly.com)](https://plotly.com/r/line-charts/)

```{r plotly-clean-NW, message=FALSE, warning=FALSE}
NW <- read_csv("data/interactive_bulletin_charts_agecl_median.csv") %>%
    select(Category, year, Net_Worth)  %>%
    group_by(Category) %>%
    arrange(year) %>%
    ungroup()
```

如果希望滑鼠在移到折線上時就會有浮出的資訊（tips）顯示該資料點的屬性特徵，可以採用`plotly()`這個套件。這個套件原本就是做線上互動圖表的，但他開發了R client讓R的使用者可以很輕易地把ggplot2的結果轉為互動圖表。但這所謂的互動也僅限於滑鼠移過去所浮出的資訊罷了，不過已經能夠達到吸引部分讀者目光、提供訊息的效果。

而plotly的設計非常簡單，就是把ggplot的結果指給一個變數後，然後用`ggplotly(NW.plot)`將其轉為plotly即可。但要注意的是，並不是每一個圖都可以順利轉換。例如本節最後一個例子Treemap便無法成功轉換。

設定：原本plotly會帶一個操控列，可以在`ggplotly()`指令後加入`config()`便可將其隱藏。

```{r plotly-NW-plot1, message=FALSE, warning=FALSE}
NW.plot <- NW %>%    
    ggplot() + 
    aes(year, Net_Worth, color=Category) + 
    geom_line() + 
    theme_minimal() + 
    labs(title = "Net Worth by year grouped by age groups",
         x = "Year",
         y = "Net Worth") + th

library(plotly)
ggplotly(NW.plot) %>%
  config(displayModeBar = FALSE)
```

可以在`aes()`設定要帶入圖的變數時，新增一個`text`變數，手動設定要呈現的動態呈現方塊。但要注意的是，要多加入一個`group=1`才能夠作用（WHY？）。但前例浮出視窗的原始內容所顯示的是原本的變數名稱和值，往往不易觀察。比較好的方式是在下`ggplot() + aes()`指令時，在`aes()`中指定`text`來作為後續浮出視窗內容。指定方法如下。要注意的是，該浮出視窗的語法是HTML，所以如果要改寫浮出視窗內容，要用`paste0()`將變數和HTML的標籤給銜接起來。以下例子中的`<b>`代表粗體的意思，`<br>`則是換行符號。

```{r plotly-NW-plot2, message=FALSE, warning=FALSE}
NW.plot <- NW %>%    
    ggplot() + 
    aes(year, Net_Worth, 
        color=Category, 
        text = paste0("<b>年(X): </b>", year, "<br>",
                      "<b>淨資產(Y): </b>", Net_Worth,"<br>",
                      "<b>年齡組: </b>", Category),
        group=1) + 
    geom_line() + 
    theme_minimal() + 
    labs(title = "Net Worth by year grouped by age groups",
         x = "Year",
         y = "Net Worth") + th

ggplotly(NW.plot, tooltip = "text") %>%
  config(displayModeBar = FALSE)
```

其他例子中使用`ggplotly()`都是直接照前面的方法套用即可。唯獨在Treemap中無法用這樣的方法來做互動的視覺化。想想這也正常，畢竟Treemap是用非ggplot的第三方套件（`library(treemapify)`）。

除此之外，可以把R Markdown中Code Cell的的設定加入`include=FALSE`，這樣可以讓RMD在編製為HTML檔時，不要顯示程式碼，而直接顯示互動的視覺化介面。

### SCATTER

```{r plotly-clean-unicef, message=FALSE, warning=FALSE}
bw <- read_csv("data/unicef-changing-childhood-data.csv") %>% 
    select(country = WP5, age = WP22140, bw = WP22092) %>%
    mutate(country = ordered(country, 
                             levels=c(1, 3, 4, 10, 11, 12, 13, 14, 17, 
                                      29, 31, 33, 35, 36, 60, 61, 77, 
                                      79, 81, 87, 165), 
                             labels=c("USA", "Morocco", "Lebanon",
                                      "Indonesia","Bangladesh", "UK", 
                                      "France", "Germany", "Spain", 
                                      "Japan", "India", "Brazil", 
                                      "Nigeria", "Kenya", "Ethiopia",
                                      "Mali", "Ukraine", "Cameroon", 
                                      "Zimbabwe","Argentina", "Peru"))) %>%
    count(country, age, bw) %>%
    group_by(country, age) %>%
    mutate(perc = n/sum(n)) %>% 
    ungroup() %>%
    filter(bw == 1) %>%
    select(country, age, perc) %>%
    spread(age, perc) %>%
    rename(`15-24y` = `1`, `40+y` = `2`)
```

```{r  plotly-ggplot-unicef, message=FALSE, warning=FALSE}
bw.p <- bw %>%
    ggplot() + aes(`40+y`, `15-24y`, label = country) + 
    geom_point(color = "skyblue", size = 2) + 
    xlim(0.2, 0.85) + ylim(0.2, 0.85) + 
    geom_text(hjust = -0.1, vjust = -0.5) + 
    geom_abline(intercept = 0, slop = 1, 
                color="lightgrey", alpha=0.5, linetype="dashed") + 
    th + 
    theme(aspect.ratio=1)
```

```{r plotly-ggplotly, message=FALSE, warning=FALSE}
bw.p %>% ggplotly()
```

### Barplot

[Bar charts with R (plotly.com)](https://plotly.com/r/bar-charts/)

```{r plotly-clean-twpopulation, message=FALSE, warning=FALSE}
county <- read_csv("data/tw_population_opendata110N010.csv") %>%
    slice(-1, -(370:375)) %>%
    type_convert() %>%
    mutate(county = str_sub(site_id, 1, 3)) %>%
    group_by(county) %>%
    summarize(
        area = sum(area), 
        people_total = sum(people_total)
    ) %>%
    ungroup()

population.p <- county %>%
  mutate(county = reorder(county, people_total)) %>%
  ggplot() + aes(county, people_total) %>%
  geom_col(fill="skyblue") +
  coord_flip() + th
```

```{r plotly-twpopulation}
population.p %>% ggplotly()
```

### Boxplot

[Box plots with ggplot2 (plotly.com)](https://plotly.com/ggplot2/box-plots/)

```{r plotly-clean-AQI, message=FALSE, warning=FALSE}
aqi.toplot <- read_rds("https://github.com/p4css/R4CSS/raw/master/data/AQI_Chaozhou.rds") %>%
    arrange(日期)%>%
    filter(測項=="PM2.5") %>%
    gather("hour", "PM25", 4:28) %>%
    mutate(PM25 = as.numeric(PM25)) %>%
    drop_na() %>%
    mutate(year = lubridate::year(日期), month = lubridate::month(日期)) %>%
    filter(month %in% c(11, 12, 1, 2, 3))

aqi.plot <- aqi.toplot %>%
    mutate(year = as.character(year)) %>%
    ggplot() + aes(x=year, y=PM25)  + 
    geom_boxplot(fill="skyblue", alpha=0.2) + 
    ylim(0, 200) + 
    coord_flip() + 
    theme_minimal()
```

```{r plotly-AQI, message=FALSE, warning=FALSE}
aqi.plot %>% ggplotly
```

### Treemap (Global Carbon)

其他例子中使用`ggplotly()`都是直接照前面的方法套用即可。唯獨在Treemap中無法用這樣的方法來做互動的視覺化。想想這也正常，畢竟Treemap是用非ggplot的第三方套件（`library(treemapify)`）。

```{r plotly-clean-treemap, message=FALSE, warning=FALSE}
totreemap <- read_csv("data/GCB2021v34_MtCO2_flat.csv") %>% 
    drop_na(`Total`) %>%
    filter(!Country %in% c("Global", "International Transport")) %>%
    filter(Year==2020) %>%
    arrange(desc(`Total`)) %>%
    mutate(perc = Total/sum(Total)) %>%
    slice(1:20)
library(treemapify)
carbon.p <- totreemap %>%
    ggplot() + aes(area = perc, fill=`Per Capita`, label=Country) +
    geom_treemap() + 
    geom_treemap_text(color="white", 
                      place="centre", 
                      grow=TRUE
                      )
```

```{r fail-ggplotly}
# carbon.p %>% ggplotly
```