2020 Candidate Climate Leadership Scores

Climate Hawks Vote 2015 Scores.csv

climatehawks.Rmd

---
title: "R Notebook"
output:
  html_document:
    df_print: paged
---

```{r setup}
library(tidyverse)
library(plotly)
library(ggrepel)

dat <- read_csv("Climate Hawks Vote 2015 Scores.csv")
```


```{r}

dat %>%
  select_if(is.numeric) %>%
  as.matrix %>%
  cov %>%
  as.data.frame() %>% 
  rownames_to_column("var1") %>%
  gather(var2, value, -var1) %>%
  mutate(type = ifelse(var1 == var2, "var", "cov")) ->
  varcov
  
varcov %>%
  arrange(type, desc(value))
```

Component with most variance: Public Engagement

```{r}
varcov %>%
  filter(var1 == "Public Engagement", type == "cov") %>%
  arrange(value)
```


High variance components least correlated with public engagment: caucuses  

```{r}
dat %>%
  ggplot(aes(y = `Public Engagement`, x = Caucuses)) +
  geom_point()
```

Despite high vairance, Caucses is nearly dichotomous. Probbaly better to bin and place on a qualitative mapping.

```{r}
dat %>%
  ggplot(aes(y = `Public Engagement`, x = `Bills Cosponsored`)) +
  geom_point()
```

```{r}
dat %>%
  ggplot(aes(y = `Public Engagement`, x = `Bills Sponsored`)) +
  geom_point()
```

Public Engagement versus Bills **Co**sponsored gives the most varied distribution - will use these for the positional mappings. 

```{r}
bgc <- "#111111"
bg <- element_rect(color = bgc, fill = bgc)
dat2 <- dat %>%
  mutate(Caucuses =`levels<-`(
    Hmisc::cut2(Caucuses, c(0, 1, 10, 11)),
    c("None", "Some", "All")
  ))
base_plot <- dat2 %>%
  ggplot(aes(y = `Public Engagement`, x = `Bills Cosponsored`,
             label = Name)) +
  geom_point(aes(color = Website, size = `Bills Sponsored`,
                 shape = Caucuses),
             # position = position_jitter(0.1, 0.1, 100),
             fill = "grey80", stroke = 1.5) +
  scale_shape_manual("Caucus\nMembership", values = c(1, 21, 19)) +
  scale_color_gradientn("Website Score",
                        colors = RColorBrewer::brewer.pal(5, "RdYlBu"),
                        limits = c(-10, 10)) +
  labs(
    x = "Bill Cosponsorship Score", y = "Public Engagement Score",
    title = "Climate Leadership Scores",
    subtitle = "for potential 2020 Democratic candidates",
    caption = "data from Climate Hawks Vote 2015 Senate scorecards"
  ) +
  theme(
    panel.background = bg,
    text = element_text(color = "white"),
    plot.background = bg,
    legend.background = bg,
    legend.key =  bg,
    panel.grid = element_blank(),
    legend.margin = margin(0, 0, 0, 0),
    axis.text = element_text(color = "white"),
    axis.ticks = element_line(color = "white")
  )
static_plot <- base_plot +
  geom_label_repel(data = filter(dat, `2020 Hopeful` == "Yes"),
                   # position = position_jitter(0.1, 0.1, 100),
                   hjust = 1.15, vjust = "middle", alpha = 0.9,
                   segment.color = "grey80",
                   direction="y", seed = 100) +
  scale_size("Bill Sponsorship Score", limits = c(-5, 15),
             breaks = c(-5, 5, 15)) +
  guides(
    shape = guide_legend(
      override.aes = list(color = c("#ffffbf"), size = 4)
    ),
    size = guide_legend(
      override.aes = list(color = c("#ffffbf"))
    )
  ) +
  theme(text = element_text(size = 16, color = "white"))
static_plot
```

```{r}
size <- 5
ggsave("climate scores.png", static_plot, height = size, width = size * 1.618)
```

```{r}
# fix legend label being placed incorrectly - remove and add manually
interactive_plot <- plotly::ggplotly(base_plot +
                   guides(size = guide_legend(""),
                          shape = guide_legend(""))) %>%
  add_annotations(text="Caucus\nMembership",
                  xref="paper", yref="paper",
                  x=1.03, xanchor="left",
                  y=0.35, yanchor="bottom",
                  legendtitle=TRUE, showarrow=FALSE)
# remove excess padding
interactive_plot$sizingPolicy$padding <- "0"
# highlight 2020 contenders
# (can't use data index because data is split by symbol)
c2020 <- dat$Name[dat$`2020 Hopeful` == "Yes"] %>%
  paste(collapse = "|")
interactive_plot$x$data <- lapply(interactive_plot$x$data, function (trc) {
  if (length(trc$text) == 0) { return(trc) }
  sel <- grep(c2020, trc$text)
  trc$selectedpoints <- sel - 1 # zero-based js indexing
  trc
})
htmlwidgets::saveWidget(
  interactive_plot, "index.html", libdir = "lib",
  title = "2020 Candidate Climate Leadership Scores",
  selfcontained = FALSE, background = bgc
)
interactive_plot
```

Abandoned concepts and other scratch work below:

```{r}
rank_order <- dat$Name[order(dat$Total)]
dat %>%
  ggplot(aes(y = `Public Engagement`, x = Name,
             color = `2020 Hopeful`)) +
  geom_point() +
  scale_x_discrete(limits = rank_order) +
  coord_flip()
```


```{r}
table(dat$`Bills Sponsored`)
```

```{r}
table(dat$`Website`)
```

```{r}
Hmisc::cut2(dat$Caucuses, c(0, 1, 10, 11)) %>%
  table
  
```