phd day workshop_examples.Rmd

---
title: "Beyond barcharts"
output: html_notebook
---

# Bar chart alternatives

Let's meet the iris dataset!
Read more about it in the R help page for iris

```{r}
head(iris2)
```

## ggplot basics

Let's do ggplot basics now!

Make a jitterplot with `geom_jitter()` Set the `x aesthetic` to `Species` and the `y aesthetic` to `Sepal.Length`

```{r}
ggplot(iris,mapping = aes(y = Sepal.Length,x=Species)) +
  geom_jitter()
```

## Barchart in ggplot

Let's make the same plot, but as a barchart.
Do you already know `stat_summary()`?
It's ggplot's best kept secret.
Replace `geom_jitter()` with `stat_summary(geom = 'bar')` and make a barchart.

Second, add `stat_summary(geom = 'errorbar')`

```{r}
ggplot(iris,mapping = aes(y = Sepal.Length,x=Species)) +
  stat_summary(geom = 'bar') +
  stat_summary(geom = 'errorbar')
```

That's your barchart right there!
Is it a beauty, no not quite, but we can fine-tune that in a later stage.

## Density alternatives

Let's make a couple alternatives.

### try `geom_violin()`

```{r}
ggplot(iris,mapping = aes(y = Sepal.Length,x=Species)) +
   geom_violin()
```

### try `geom_density`

Replace the `x` aesthetic for `col = Species` , and maybe swap x and y axes.

You can add the `linetype` aesthetic as well!
that way it works in greyscale print.

```{r}
ggplot(iris,mapping = aes(x = Sepal.Length,col=Species,linetype = Species)) +
  geom_density()
```

-------====== Wait here please, we'll plenary get up to speed in a second ===========------

## point/count alternatives

### dotplot

Similar to a violin plot, but shows the actual data points.
Use `geom_dotplot(binaxis = 'y',stackdir = 'center')`

```{r}
ggplot(iris,mapping = aes(y = Sepal.Length,x= Species)) +
  geom_dotplot(binaxis = 'y',stackdir = 'center')
```

### histogram

```{r}
ggplot(iris,mapping = aes(x = Sepal.Length,fill= Species)) +
  geom_histogram(position = 'identity',alpha=.6)
```

### boxplot

```{r}
ggplot(iris,mapping = aes(y = Sepal.Length,x= Species)) +
  geom_boxplot()
```

-------====== Wait here please, we'll plenary get up to speed in a second ===========------

# Combining shapes!

ggplot can layer shapes on top of one another.
We can use this to our advantage, and add more data in our figure.

### Boxplot with jitter

Make a boxplot like before, but add the datapoints with `geom_jitter`.
Make the datapoints transparent and grey if you know how.

```{r}
ggplot(iris,mapping = aes(y = Sepal.Length,x= Species)) +
  geom_boxplot() +
  geom_jitter(alpha = .5, colour = 'grey', width = .2) +
  theme_classic()
```

### dotplot with median

Was that too easy, then here is a hard one!

Next, make a dotplot, but add a red line that displays the median.
Use stat summary and the cheatsheet.

*hint: `stat_summary()` has a `fun=` argument for determining what function you want used to summarise the data.*

```{r}
ggplot(iris,mapping = aes(y = Sepal.Length,x= Species)) +
  geom_dotplot(binaxis = 'y',stackdir = 'center') +
  stat_summary(geom = 'crossbar',fun=median,col = 'red') + 
  theme_classic()
```

-------====== Wait here please, we'll plenary get up to speed in a second ===========------

# Extension packages

## GGdist

ggdist, deals with displaying distributions, perfect!
Have a look at their cheatsheet.
[ggdist cheatsheet](https://github.com/mjskay/ggdist/blob/master/figures-source/cheat_sheet-slabinterval.pdf){.uri}

Try a ggdist visualisation on the iris dataset.

```{r}
ggplot(iris,mapping = aes(y = Sepal.Length,x= Species,shape = Species)) +
  ggdist::stat_halfeye() 

```

# A figure of your own?

Do you have a (barchart?) figure of your own at hand?
Maybe play with that!
I'll be around to help out.

```{r}

```

## Alternative:

Formatting your data into the right shape is as hard as working ggplot itself.
I'd like you to try to make a boxplot of the irisdataset.
On the x axis, should be the different lengths that were measured.
The fill colour should be the Species, and on the y axis the actual measured length.
Can you manage?

Now make your favourite alternative, maybe a dotplot with a crossbar?
Or some other shape you may have found in the ggdist package.

```{r}

```

## Challenge!

Feeling up to a challenge, try to re-create this:

![](images/clipboard-819503216.png)

```{r}
iris_long <- pivot_longer(iris, -Species, names_to = "variable", values_to = "value")
iris_long$variable <- factor(iris_long$variable,
                             levels = c("Petal.Width", 
                                        "Sepal.Width", 
                                        "Petal.Length" ,
                                        "Sepal.Length"))



# make a dotplot with ggplot
ggplot(iris_long, aes(x = variable, y = value, fill = variable)) +
  #geom_dotplot(binaxis = "y", stackdir = "center", position = "dodge",alpha = .5) +
  geom_violin() +
  facet_wrap(~Species, scales = "free_y") +
  stat_summary(fun = mean, geom = "pointrange", shape = 18, size = 1, color = "black",show.legend = F) +
  stat_summary(fun = mean,geom = 'line', aes(group = 1), size = 1) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))
```