Boxplots_solutions.Rmd

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

```{r, message = FALSE, echo = FALSE}
phoible <- read.csv("/Users/alena2/Library/CloudStorage/Dropbox/Teaching_Students/Classes_Courses/2023-2024/HUJI2024_Quant/Class5-6/phoible.csv")
```

```{r, message = FALSE, echo = FALSE}
library(tidyverse)
library(readxl)
require(forcats)
library(cowplot)
library(knitr)
library(kableExtra)
```

```{r, message = FALSE}
head(phoible)
```

# Task 1: Indo-European (ieur) vs. Nilo-Saharan (nsah)
* Consider the parallel boxplots of the Indo-European (ieur) and Nilo-Saharan (nsah) language families.

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% 
        filter(fam %in% c("nsah", "ieur")) %>%
        ggplot(aes(y = phonemes, x = fam, fill = fam)) +
        geom_boxplot()+
        labs(title="Number of phonemes in Indo-European and Nilo-Saharan languages",
             x ="family", y = "number of phonemes") +
        theme_light()+
        theme(axis.text.y = element_text(size = 12), axis.text.x = element_text(size = 12))
```

* Answer the following questions:
1. Which distribution has outliers? How many?
2. Which distribution is more compact?
3. Which distribution has the largest range (with and without the outliers)?
4. What is the range of Nilo-Saharan phoneme inventories?
5. Which distribution is negatively skewed?
6. How many languages are in each boxplot?

## Solutions

6. How many languages are in each boxplot? 
One cannot say ont he basis of the boxplot, but one can always add individual datapoints:

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% 
        filter(fam %in% c("nsah", "ieur")) %>%
        ggplot(aes(y = phonemes, x = fam, fill = fam)) +
        geom_boxplot()+
        labs(title="Number of phonemes in Indo-European and Nilo-Saharan languages",
             x ="family", y = "number of phonemes") +
        theme_light()+
        theme(axis.text.y = element_text(size = 12), axis.text.x = element_text(size = 12))
```

# Task 2
* Consider the parallel boxplots of four language families:

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% 
        filter(genus %in% c("Mayan", "Germanic", "Semitic", "Indic")) %>%
        ggplot(aes(y = con, x = genus, fill = genus)) +
        geom_boxplot()+
        labs(title="Number of consonants in four families",
             x ="family", y = "number of consonants") +
        theme_light()+
        theme(axis.text.y = element_text(size = 12), axis.text.x = element_text(size = 12))
```

* The data set (number of consonants) for one of the four languages families is this:

```{r, message = FALSE, echo = FALSE}
phoible %>% filter(genus == "Germanic") %>% select(con) -> con
kable(con) %>%
  kable_minimal(full_width = F)
```

* Which boxplot corresponds to this dataset? Discuss, the various aspects of the boxplot which make the identification possible.

## Solutions

* Germanic
```{r, message = FALSE, echo = FALSE}
phoible %>% filter(genus == "Germanic") %>% select(language, con) -> con
kable(con) %>% 
  kable_minimal(full_width = F)
```

# Task 3: boxplot vs. density plot
* Consider the parallel boxplots of the number of phonemes in four language families.
* One of the language families is visualized with a density plot. Which one is it?

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% 
        filter(fam %in% c("nsah", "sitb", "ieur", "drav")) %>%
        ggplot(aes(y = phonemes, x = fam, fill = fam)) +
        geom_boxplot()+
        labs(title="Number of phonemes in four families") +
        theme_light()+
        theme(axis.text.y = element_text(size = 12), axis.text.x = element_text(size = 12))
```

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% filter(fam %in% c("nsah")) %>% 
  ggplot(aes(x = phonemes)) +
  geom_density(fill = "red", alpha = 0.3) + 
  labs(title="Number of phonemes in a language family") +
  theme_light()+
  theme(axis.text.y = element_text(size = 12), axis.text.x = element_text(size = 12))
```

## Solutions
* Nilo-Saharan

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% filter(fam %in% c("nsah")) %>% 
  ggplot(aes(x = phonemes)) +
  geom_density(fill = "red", alpha = 0.3) + 
  labs(title="Number of phonemes in the Nilo-Saharan languages") +
  theme_light()+
  theme(axis.text.y = element_text(size = 12), axis.text.x = element_text(size = 12))
```

# Task 4: boxplot vs. density plot
* Consider the parallel boxplots of the number of phonemes in four language families.
* Two of the language families are visualized with the density plots. Which oness are they?
* Explain how you can figure it out.

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% 
        filter(fam %in% c("ausa", "anes", "aust", "ieur")) %>%
        ggplot(aes(y = phonemes, x = fam, fill = fam)) +
        geom_boxplot()+
        labs(title="Number of phonemes in four families",
             x ="family", y = "number of phonemes") +
        theme_light()+
        theme(axis.text.y = element_text(size = 12), axis.text.x = element_text(size = 12))
```

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% filter(fam %in% c("aust", "anes")) %>% 
  ggplot(aes(x = phonemes, fill = fam)) +
  geom_density(alpha = 0.3) + 
  labs(title="Number of phonemes in two language families") +
  theme_light()+
  theme(axis.text.y = element_text(size = 14), axis.text.x = element_text(size = 14)) +
  theme(legend.position="none")
```

## Solutions

Australian (`aust`) and Austronesian (`anes`)

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% filter(fam %in% c("aust", "anes")) %>% 
  ggplot(aes(x = phonemes, fill = fam)) +
  geom_density(alpha = 0.3) + 
  labs(title="Number of phonemes in two language families") +
  theme_light()+
  theme(axis.text.y = element_text(size = 14), axis.text.x = element_text(size = 14))
```

# Task 5: boxplot vs. histogram
* Consider the parallel boxplots of the number of phonemes in four language families.
* Three of the language families are visualized with the histograms. Which one is missing?
* Explain how you can figure it out.

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% 
        filter(fam %in% c("ausa", "altc", "trng", "anes")) %>%
        ggplot(aes(y = phonemes, x = fam, fill = fam)) +
        geom_boxplot()+
        labs(title="Number of phonemes in four language families",
             x ="family", y = "number of phonemes") +
        theme_light()+
        theme(axis.text.y = element_text(size = 12), axis.text.x = element_text(size = 12))
```

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% filter(fam %in% c("ausa", "altc", "trng")) %>% 
  # change order of factor by the family size
  mutate(fam = fct_infreq(fam)) %>% 
  ggplot(aes(x = phonemes, fill = fam)) +
  geom_histogram(binwidth = 5) + 
  labs(title="Number of phonemes in three language families",
  x ="family", y = "number of phonemes") +
  theme_light()+
  theme(axis.text.y = element_text(size = 14), axis.text.x = element_text(size = 14)) +
  theme(legend.position="none") + 
  facet_grid(. ~ fam ) +
  theme(strip.text.x = element_text(size=0))
```

## Solutions

* The missing family is Austronesian (`anes`)

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% filter(fam %in% c("ausa", "altc", "trng")) %>% 
  # change order of factor by the family size
  mutate(fam = fct_infreq(fam)) %>% 
  ggplot(aes(x = phonemes, fill = fam)) +
  geom_histogram(binwidth = 5) + 
  labs(title="Number of phonemes in three language families",
  x ="family", y = "number of phonemes") +
  theme_light()+
  theme(axis.text.y = element_text(size = 14), axis.text.x = element_text(size = 14)) +
  theme(legend.position="none") + 
  facet_grid(. ~ fam )
```

```{r, message = FALSE, echo = FALSE, out.width="60%"}
phoible %>% filter(fam %in% c("ausa", "altc", "trng", "anes")) %>%
  ggplot(aes(x = phonemes, fill = fam)) +
  geom_histogram(binwidth = 5) + 
  labs(title="Number of phonemes in four language families") +
  theme_light()+
  theme(axis.text.y = element_text(size = 14), axis.text.x = element_text(size = 14)) +
  theme(legend.position="none") + 
  facet_grid(. ~ fam)
```