rladies_paris_ggplot.Rmd

---
title: "R-Ladies intro to ggplot2 tutorial"
author: "Sarah Hosking"
date: "March 7, 2017"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)

library(lubridate)
library(tidyverse)
library(GGally)

library(reshape2) #plot missing vals

theme_set(theme_bw())

```

# Paris Air Pollution

Because we all need a break from car, iris and most of all, election, data.

## Load Data

```{r}

airparif <- read_rds('airparif.rds')
dim(airparif)
```

```{r}
str(airparif)
```

```{r}
head(airparif)
```

```{r}
tail(airparif)
```

```{r}
summary(airparif)
```


## What are we looking at?


```{r}

names(airparif)

```

What the abbreviations mean:

* `PM25` = fine particulate < 2.5 mm
* `PM10` = fine particulate < 10 mm
* `03` = Ozone
* `NO2` = Nitrogen Dioxide (Azote)
* `CO` = Carbon monoxide

## Explore the data


### Plot a correlation matrix

```{r ggpairs}

# create a sample - ggpairs takes time.
set.seed(888)
sample <- subset(airparif, select = c(PM10, PM25, NO2, O3, month))
sample <- sample[sample(1:nrow(sample),1000),]

#ggpairs()
```


## Summarize and plot

First, need to prepare the data

```{r}

# Option 1: load the summarized data
#airparif.long <- read_rds("airparif_long.rds")


# Option 2: summarize data yourself
# airparif.long <- airparif %>% 
#   gather(, 
#          key = , value = , 
#          na.rm = TRUE) %>% 
#   group_by() %>% 
#   summarise(mean = ,
#             median = ,
#             max =)

```

Now plot this new df

```{r}

```

## How many days at each pollution level?

```{r}

# Option 1: load the summarized data
#levels <- read_rds("levels.rds")


# Option 2: summarize data yourself

# First, load function to tag pollution levels
pollution.level <- function(x, lower = 0, upper = 100, 
                            by = 25, sep = "-", above.char = "+") {
 
 labs <- c("very low", "low", "medium", "high", "very high")

 cut(floor(x), 
     breaks = c(seq(lower, upper, by = by), Inf),
     right = FALSE, 
     labels = labs)
}

# Then complete this code
# levels <-  airparif.long %>%
#   group_by( ) %>% 
#   mutate_all(funs())

```


How many days at each level for each pollutant?

```{r}

# create a bar plot
# p <- ggplot(levels, aes(median, fill = pollutant))
# p + geom_bar()

```

Make this easier to read

```{r}

```


# BONUS STUFF (If we have time)

## Make plots prettier

```{r}

# only plot PM pollutants
# particles <- airparif.long %>% 
#   filter(pollutant %in% c('PM10', 'PM25')) 
# 
# 
# p <- ggplot(particles, aes(date, mean))
# 
# p + 
#   geom_line(aes(y = median, colour = pollutant)) + 
#   scale_x_date() + 
#   theme_
  


```


## qplot

`qplot` is part of `ggplot2`, and stands for "quick plots". Its syntax is similar to base plotting, and is less verbose than `ggplot()`.

```{r qplot hists}

qplot(PM10, data = airparif)

# increase bins
qplot(data = airparif, PM10, binwidth = 5)

```


It's meant, however, to work with vectors. Here's a demo.

```{r}

# qplot meant to work with vectors
a <- c('A', 'B', 'C')
b <- c(1, 2, 3)

# qplot
q <- qplot(a,b)
q

```


```{r}

# ggplot needs a dataframe
d <- data.frame(a,b)

# ggplot
p <- ggplot(data = d, aes(a,b)) +
  geom_point()
p


```


```{r}

# change var
a <- c('X','Y','Z')

# with qplot
q
```

```{r}

# with ggplot
p
```

## Visualize missing values with ggplot

```{r}
# credit to:
# http://www.njtierney.com/r/missing%20data/rbloggers/2015/12/01/ggplot-missing-data/
missing.data <- function(x){
  
  x %>% 
    is.na %>%
    melt %>%
    ggplot(data = .,
           aes(x = Var2,
               y = Var1)) +
    geom_raster(aes(fill = value)) +
    scale_fill_grey(name = "",
                    labels = c("Present","Missing")) +
    theme_minimal() + 
    theme(axis.text.x  = element_text(angle=45, vjust=0.5)) + 
    labs(x = "Variables in Dataset",
         y = "Rows / observations")
}

#missing.data(airparif[,1:7])
```