---
title: "Lab 13: Grammar of Graphics"
output: html_document
---


You'll need one new package today, something I wrote called smodels:

```{r}
devtools::install_github("statsmaths/smodels")
```

Then, read in the standard libraries.

```{r}
knitr::opts_chunk$set(echo = TRUE)
library(readr)
library(ggplot2)
library(dplyr)
library(smodels)
theme_set(theme_minimal())
```

## Tea Reviews

Here, we will take look at a dataset of tea reviews from Adagio Teas:

```{r}
tea <- read_csv("https://statsmaths.github.io/stat_data/tea.csv")
```

With the following variables:

- name: the full name of the tea
- type: the type of tea. One of:
    - black
    - chai
    - decaf
    - flavors
    - green
    - herbal
    - masters
    - matcha
    - oolong
    - pu_erh
    - rooibos
    - white
- score: user rated score; from 0 to 100
- price: estimated price of one cup of tea in cents
- num_reviews: total number of online reviews

Draw a scatter plot with num_reviews (x-axis) against score (y-axis):

```{r}

```

Now add a best fit line to the scatter plot:

```{r}

```


Does the score tend to increase, decrease, or remain the same as the number
of reviews increases?

**Answer**:

Create a text plot with score (x-axis) against price (y-axis) using the tea
name as a label. What is the most expensive tea in the data?

**Answer**:

Which tea has the lowest score? Does it have a high price per cup?

**Answer**:

Create a scatter plot of with num_reviews (x-axis) against score (y-axis) and
a facet by tea type:

```{r}

```

Repeat the above plot but allow the scales to be float freely:


```{r}

```

Which of the two facet plots do you find more readable and why? (Hint: there is
not a right answer here)

**Answer**:

## Exploratory Analysis of Car Speeding Data

To see some of the commands that we will be looking at next time, let's grab
one more dataset. The data here concerns a study that measured the speed at
which cars were traveling. From the documentation:

> In a study into the effect that warning signs have on speeding
> patterns, Cambridgeshire County Council considered 14 pairs of
> locations. The locations were paired to account for factors such
> as traffic volume and type of road.  One site in each pair had a
> sign erected warning of the dangers of speeding and asking drivers
> to slow down.  No action was taken at the second site.  Three sets
> of measurements were taken at each site.  Each set of measurements
> was nominally of the speeds of 100 cars but not all sites have
> exactly 100 measurements. These speed measurements were taken
> before the erection of the sign, shortly after the erection of the
> sign, and again after the sign had been in place for some time.

To read the dataset in, run the following line of code (if you get an
error, you probably forgot to run the code at the top of the file):

```{r}
amis <- read_csv("https://raw.githubusercontent.com/statsmaths/stat_data/gh-pages/amis.csv")
```

Open the dataset in the data viewer. You'll see four variables: the
observed speed of the car, the period of the observation (whether it was
before signs were put up, right after, or far after they were put up),
a variable warning indicating whether a location had a sign erected,
and "location" indicating one of 14 location pairs.

The following command plots the average value for each period colored by
whether there was a warning given or not:

```{r}
ggplot(amis, aes(period, speed)) +
  geom_mean(aes(color = warning))
```

How would you explain the observed effects in words? Does it seem from
this plot that the signs had their indented effect of slowing traffic?

**Answer**:

The following command plots the average speed at each pair of locations
based on whether there was a warning sign or not **in the before period*.
That is, these measurements come before any signs were put in place.

```{r}
ggplot(filter(amis, period == "Before"), aes(location, speed)) +
  geom_mean(aes(color = warning))
```

The study design suggests that the pairs of locations indicate similar
road conditions. Does this seem correct to you based on this plot? What
locations seem particularly suspicious?

**Answer**: