Geospatial Analysis with R

class: center, middle, inverse, title-slide

.title[
# Geospatial Analysis with R
]
.subtitle[
## Class 13
]

---

## Election Results
<img src="class13_files/figure-html/unnamed-chunk-1-1.png" style="display: block; margin: auto;" />

---

```r
library(tidyverse)
library(rvest)

URL <- "https://history.house.gov/Institution/Party-Divisions/Party-Divisions/"
webpage <- read_html(URL)
cong <- as_tibble(html_table(webpage)[[1]])
onms <- colnames(cong)
newnms <- c("congress", "seats", "D", "R", "other", "delres")
cong <- cong %>% 
  slice((which(grepl("^Republican", `Anti-Administration`))[1] + 1):nrow(.)) %>%
  filter(`Congress (Years)` != colnames(cong)[1]) %>% 
  mutate(year = gsub("(*.*-)|(*.*–)|)|)2", "", `Congress (Years)`)) %>%
  mutate(year = as.numeric(year) - 2) %>% 
  rename_at(vars(onms), ~newnms) %>% select(-other, -delres) %>%
  mutate(seats = substr(seats, 1, 3)) %>% 
  mutate_at(.vars = vars(seats, D, R), as.numeric) %>% 
  mutate(swing = (D - R) / (D + R)) 
cong %>%  
  ggplot() + geom_line(aes(year, swing)) + 
  theme(axis.text.x = element_text(angle = 90, vjust = 0.5)) + 
  geom_hline(yintercept = 0, lty = 2, col = "red") + 
  xlab("") + ylab("(D - R) / (D + R)") + 
  ggtitle("Normalized Party Control Index for US House") + theme_linedraw() + 
  scale_x_continuous(breaks = seq(1859, 2019, 10), expand = c(0, 2))
```

---
## Working with data, continued

### `do.call` / `bind_rows`

```r
sz <- 100
for(i in 1:3) {
  set.seed(i)
  d <- data.frame(
    id = 1:sz,
    v1 = runif(sz, min = 2, max = 12),
    grp = paste0("g", i)
  ) %>% mutate(v2 = v1 + rnorm(sz, mean = 2, sd = 2)) %>% 
    select(id, v1, v2, grp)
  readr::write_csv(d, file = file.path(tempdir(), paste0("dataset", i, ".csv")))
}
# ggplot(d) + geom_point(aes(x = v1, y = v2))

fs <- list.files(tempdir(), pattern = "dataset.*.csv", full.names = TRUE)
# dat <- do.call(rbind, lapply(fs, readr::read_csv))
dat <- bind_rows(lapply(fs, readr::read_csv))

plot(dat$v1, dat$v2)
```

---
### Manipulating and analyzing data

- reshape
- mutate
- select
- joins
- split-apply-combine
- plotting
- regression

---

## Reshape

```r
dat %>% 
  pivot_wider(names_from = grp, values_from = v1:v2)

dat %>% 
  select(-v1) %>%
  pivot_wider(names_from = grp, values_from = v2)

dat_wide <- dat %>% 
  select(id, v2, grp) %>%
  pivot_wider(names_from = grp, values_from = v2)

dat_long <- dat_wide %>% 
  pivot_longer(g1:g3, names_to = "grp", values_to = "v2") %>% 
  arrange(grp)

dat_long2 <- dat_wide %>% 
  pivot_longer(g1:g3, names_to = "grp", values_to = "v2")

cbind(dat, dat_long) %>% head()
bind_cols(dat, dat_long) 
```

---
## Joins

```r
cbind(dat, dat_long2) %>% head()
# bind_cols(dat, dat_long)

dat %>% left_join(dat_long2)
dat_long2 %>% 
  rename(grp2 = grp) %>%
  select(id, grp2, v2) %>% 
  left_join(dat, ., by = c("id", "v2", "grp" = "grp2"))

dat_long2 %>% 
  rename(grp2 = grp) %>%
  select(id, grp2, v2) %>% 
  left_join(dat %>% select(-v2), ., by = c("id", "grp" = "grp2")) %>% 
  select(id:v1, v2, grp)
```

- Note: understand the differences between `full_join`, `inner_join`, `right_join`, and `left_join`.