exercises.Rmd

---
output: github_document
---

<!-- Note: edit the .Rmd file not the .md file -->

## 1.1 R and RStudio

### 1.1a

1. Get into groups of 2 and, in those groups:
1. Create a new RStudio project
1. Create a new blank R script
1. Identify and interact with each of the 4 panels in RStudio
1. Get help on the plot function with `?plot`
1. Create a plot using the `plot()` function
1. Find and install a new package on a topic of your choice with `Tools > Install Packages` (requires internet)
1. Attach the package using `library()`
1. Find and install a new package with `install.packages()`
1. In your source panel write code that creates vector objects `x` and `y` and plots them with `plot(x, y)` to create something that looks like this:
<!-- (is it reproducible?) -->

```{r}
# hint: create a vector object of the numbers 1, 2, 3 and 6 and call it x:
x = c(1, 2, 3, 6)
```

```{r, echo=FALSE}
y = x^2
plot(x, y)
```

- Bonus: find out exactly what R version you are using (tip: use a search engine!)

- Bonus: use R to find out how many minutes you've been alive for. Feel free to use an invented age. Tip: try using both 'base' `as.POSIXct()` and 'tidyverse' `ymd_hm()` functions - you may also need to search online for this.
    
```{r, eval=FALSE, echo=FALSE}
date_of_birth1 = as.POSIXct("1985-12-20 12:00")
date_of_birth2 = ISOdatetime(
  year = 1985,
  month = 12,
  day = 20,
  hour = 12,
  min = 0,
  sec = 0
)
date_of_birth3 = lubridate::ymd_hm("1985-12-20 12:00")
current_time = Sys.time()
diff_days = current_time - date_of_birth1
diff_days * 24 * 60
difftime(current_time, date_of_birth1, units = "m")
```


```{r, echo=FALSE, eval=FALSE}
devtools::install_github("ITSLeeds/stats19")
stats19::dl_stats19_2017_ac()
ac17 = stats19::read_stats19_2005_2014_ac(data_dir = "dftRoadSafetyData_Accidents_2017", filename = "Acc.csv")
ac17 = stats19::format_stats19_2016_ac(ac17)
ac17
View(ac17)
ac_wy = ac17[ac17$Police_Force == "West Yorkshire", ]
ac_wy = dplyr::filter(ac17, Police_)
library(tidyverse)
ac_wy = ac17 %>% 
  filter(Police_Force == "West Yorkshire")
plot(ac_wy$Longitude, ac_wy$Latitude)
```

### 1.1b R classes

1. What class is each of these objects:
    ```{r}
    x = 1:6
    y = sqrt(x)
    z = y + 0.1
    z[3] = "hello"
    ```
    ```{r, echo=FALSE, eval=FALSE}
    class(x)
    class(y)
    class(z)
    typeof(x)
    is.vector(x)
    data.frame(x, y, z)
    m = cbind(x, y, z)
    class(m)
    typeof(m)
    ```
1. Create a data frame that contains variables `x`, `y` and `z` and write it out as a `.csv` file.
    - Bonus create a matrix composed of `x`, `y` and `z` variables. What type does it have?
1. Download and read-in the `ac_wy.csv` dataset using `read_csv()`. Hint, the following command may help get it:
    ```{r, eval=FALSE}
    f = "https://github.com/ITSLeeds/highways-course/releases/download/0.2/ac_wy.csv"
    download.file(url = f, destfile = "ac_wy.csv")
    ```
1. How many rows are in the `ac_wy` data frame?

1. How many lists are in the `ac_wy` data frame?

```{r, echo=FALSE}
# to be continued...
```

## 2 Stats refresher and packages

1. Discuss in groups: what kind of statistical analysis do you you do, exploratory or hypothesis testing?
1. Use `sessionInfo()` to find out what which packages are currently attached in your R session.
    - How many are there?
    - Run the command `devtools::session_info()`. What's different about the result?
1. Attach the tidyverse package. What does each of the messages mean?:
    ```{r, echo=FALSE}
    library(tidyverse)
    ```
1. How many packages are now attached?
1. Restart your R session and load some **tidyverse** packages individually. Start with **readr**, **dplyr** and **ggplot2**.
1.  Run `ggplot(data = mpg)`. What do you see?
1.  What does the `drv` variable describe?  Read the help for `?mpg` to find
    out.
1.  Make a scatterplot of `hwy` vs `cyl`.
1. Create a barplot showing the number and proportion of crashes in the `ac_wy` dataset on different types of roads using **ggplot2**:
  - Roads with different speed limits (absolute counts and proportions)
  - Different road types (A roads, B roads etc)
    ```{r, echo=FALSE, message=FALSE}
    ac_wy = readr::read_csv("ac_wy.csv")
    ggplot(ac_wy, aes(Speed_limit)) + geom_bar()
    # ggplot(ac_wy, aes(Speed_limit)) +
    #   geom_bar(mapping = aes(y = ..prop.. * 100)) +
    #   ylab("Percentage of roads")
    # ggplot(ac_wy, aes(`1st_Road_Class`)) +
    #   geom_bar() 
    # speed_table = table(ac_wy$`1st_Road_Class`)
    # speed_proportional = speed_table / nrow(ac_wy) * 100
    # speed_table_df = as.data.frame(speed_proportional) 
    # ggplot(speed_table_df, aes(Var1, Freq)) +
    #   geom_bar(stat = "identity")
    ``` 
  
# 3 Spatial data

1. Practical: [Section 3.2 to 3.2.2](https://geocompr.robinlovelace.net/attr.html#vector-attribute-manipulation) of handouts

1. Work through exercises 1:3 in the hand-outs
1. How many states:
    - Contain the letter E (hint: `?grepl`)?
    - Start with A (hint: search for "regex starts with")?
1. Plot all the 'E' states in red and plot the 'A' states with a thick border.
    ```{r, message=FALSE, echo=FALSE}
    library(sf)
    library(spData)
    library(dplyr)
    us_e = us_states %>% 
      filter(grepl(pattern = "e", x = NAME, ignore.case = T))
    us_a = us_states %>% 
      filter(grepl(pattern = "^A", x = NAME))
    plot(us_states$geometry)
    plot(us_e$geometry, col = "red", add = TRUE)
    plot(us_a$geometry, lwd = 5, add = TRUE)
    ```
- Bonus: use the `world` dataset to find the top 3 smallest and largest countries in the world by population and area, and plot them on a single map.

## 4 Spatial data II and roadworks

1. Practical: working-through sections [3.2.3 to 3.2.4](https://geocompr.robinlovelace.net/attr.html#vector-attribute-joining) of hand-outs
1. [Exercises](https://geocompr.robinlovelace.net/attr.html#exercises-1): 4 to 6 onwards
1. Identify the states that grew most and least from 2010 to 2015 and plot them.
    ```{r, message=FALSE, echo=FALSE}
    library(tmap)
    us_states$change = us_states$total_pop_15 - us_states$total_pop_10
    us_states$growth = NA
    top3 = tail(order(us_states$change), 3)
    bot3 = head(order(us_states$change), 3)
    us_states$growth[top3] = "Most"
    us_states$growth[bot3] = "Least"
    tm_shape(us_states) +
      tm_polygons("growth")
    ```
1. Create an obect that represents the boundary of the USA.
- Bonus: create a random distribution of 1000 points across the United States and subset those that are in Texas. Plot the results to check your code works, does it look something like this?
    ```{r, message=FALSE, echo=FALSE}
    us_union = st_union(us_states)
    xran = runif(n = 1000, min = st_bbox(us_states)[1], max = st_bbox(us_states)[3])
    yran = runif(n = 1000, min = st_bbox(us_states)[2], max = st_bbox(us_states)[4])
    mran_sample = st_sample(us_states, 1000)
    mran = data.frame(xran, yran, n = 1:1000)
    crs_us = st_crs(us_states)
    mran_sf = st_as_sf(mran, coords = c("xran", "yran"), crs = crs_us)
    # st_crs(mran_sample)
    mran_sf = st_sf(geometry = mran_sample)
    tex = us_states %>% 
      filter(NAME == "Texas")
    sel_tex = st_within(mran_sf, tex)
    sel_tex_binary = lengths(sel_tex) > 0
    mran_sf$in_texas = sel_tex_binary
    tm_shape(us_union) +
      tm_polygons() +
      tm_shape(mran_sf) +
      tm_dots(col = "in_texas", size = 0.2)
    ```
- Advanced option 1: Section [4.2 - Spatial operations on vector data](https://geocompr.robinlovelace.net/spatial-operations.html#spatial-vec) of Geocomputation with R
- Advanced option 2: Install the **roadworksUK** package and identify which MSOA in Ashford had the highest number of gas-related roadworks in the `htdd_ashford` dataset.