index.Rmd

---
title: "Data science for everyone"
output:
  html_document:
    theme: readable
    progressive: true
    df_print: paged
    highlight: tango
    css: css/style-all.css
runtime: shiny_prerendered
---

```{r setup, include=FALSE}
library(learnr)
library(magick)
library(dplyr)
library(magrittr)
library(readr)
library(glue)
library(purrr)
library(iNZightPlots)
library(htmltools)

knitr::opts_chunk$set(echo = FALSE, message = FALSE, warning = FALSE)

# header
tags$head(tags$link(rel = "shortcut icon", href = "https://annafergusson.com/logo.png"))

# data sources
photo_data <- read_csv("www/dogsvscats1.csv")
photo_links <- read_csv("www/dogsvscats2.csv")
photos_mini <- glue("{1:90}.png")
photos_gs <- glue("images/minigs/{photos_mini}") %>%
  image_read()

# functions
shade = function(n, .x = 0:255 / 255) { 
  rgb(.x, .x, .x)
}

mypars = inzpar(bg = "#FBEB7D", 
                col.pt="#000000", 
                fill.pt="#000000", 
                pch=20, 
                box.fill="transparent", 
                cex.dotpt = 2, 
                bar.fill = "#666666", 
                col.fun = shade, 
                cex.text = 3, 
                cex.axis = 2, 
                cex.lab = 2,
                cex.main = 2)


gs_pixels <- function(photo_num, num_pixels){
  image <- photos_gs[photo_num]
  # take a random sample of 500 pixels
  as.numeric(image_data(image)[1,,]) %>%
    sample(num_pixels)
}

grayscale_plot <- function(photo_num, num_pixels = 50){
  # take a random sample of 500 pixels
  sampledata <- data.frame(greyscale = gs_pixels(photo_num, num_pixels))
  
  # create an iNZight dot plot
  fig <- image_graph(width = 800, height = 600, res = 96)
  sampledata %>%
    iNZightPlot(greyscale, colby = factor(greyscale, levels=0:255), plottype="dot", data=., inzpars = mypars, hide.legend = TRUE, boxplot = TRUE, xlim=c(0,255), main="Greyscale plot")
dev.off()

# weird iNZight double layer plots!
fig[2] %>%
    image_annotate(paste0("mean: ", round(mean(sampledata$greyscale), 1)), size = 20, color = "black", boxcolor = "grey", location = "+5+5") %>%
    image_annotate(paste0("sd: ", round(sd(sampledata$greyscale), 1)), size = 20, color = "black", boxcolor = "grey", location = "+5+45") 
}

image_grid <- function(images, total_width = 800){
  num <- length(images)
  cols <- ceiling(sqrt(num))
  pixel <- total_width
  width <- pixel / cols 
  backing <- image_blank(width = pixel, height = pixel) %>%
    image_background("#ffffff")
  
  for(i in 1:num){
    add <- photos_gs[images[[i]]] %>% 
      image_scale(width)
    x <- (((i - 1 + cols) %% cols)) * width
    y <- (ceiling(i / cols) - 1 ) * width
    backing <- image_composite(backing, add, offset = glue("+{x}+{y}"))
  }
  backing
}

```


```{js echo=FALSE}
$(function() {
var editor;
$('.ace_editor').each(function( index ) {
  editor = ace.edit(this);
  editor.getSession().setUseWrapMode(true);
  editor.setFontSize("16px");
});
})
```

# <img src='images/logo.png' width='50' /> Data science for everyone, including cats (and maybe dogs) {.tabset}

## Check your photo-plot matches

```{r check-plots, exercise=TRUE}
grayscale_plot(photo_num = 4, 
               num_pixels = 50)
```

## Light or dark?

```{r sort-photos, exercise=TRUE, exercise.lines=20}
photo_ref <- photo_links %>%
  sample_n(10) %>%
  rowwise() %>%
  mutate(
    gs_shade = case_when(
      gs_pixels(photo_num, 500) %>% median() > 123 ~ "dark",
      gs_pixels(photo_num, 500) %>% median() < 122 ~ "light"
    )) 

print("Light photos")
photo_ref %>%
  filter(gs_shade == "light") %>%
  pull(photo_num) %>%
  image_grid()

print("Dark photos")
photo_ref %>%
  filter(gs_shade == "dark") %>%
  pull(photo_num) %>%
  image_grid()
```

## Create your base image letter

Use the drawing tool below to draw the first letter of your first name as a capital letter. 

<iframe id='drawing' src="https://script.google.com/a/aucklanduni.ac.nz/macros/s/AKfycbxLsrTYirZdR6O5kSIdSsah9FzirzCJLzPwQK0G/exec" frameborder="0" width="500" height="300" allowfullscreen="true" mozallowfullscreen="true" webkitallowfullscreen="true"></iframe>

> Copy your link below BEFORE running the code

```{r my-letter, exercise=TRUE, , exercise.lines=8}
# replace the link below with your link
my_link <- ""

# this code shows your letter in only black or white
# with no shades of gray
image_read(my_link) %>%
  image_scale("500") %>%
  image_convert(type = "bilevel")
```

## Make a simple photo mosaic!

```{r letter-mosaic, exercise=TRUE, exercise.timelimit = 180, exercise.diagnostics=FALSE, exercise.lines=13}
# replace the link below with your link
my_link <- ""

# here are the rules to label each photo
# they need your help!
photo_ref <- photo_links %>%
  rowwise() %>%
  mutate(
    gs_shade = case_when(
      gs_pixels(photo_num, 500) %>% median() > 123 ~ "dark",
      gs_pixels(photo_num, 500) %>% median() < 122 ~ "light"
    )) 

# the rest of the code makes the mosaic
# don't worry about changing this part

# read your letter drawing and convert
# to black and white only
my_letter <- image_read(my_link) %>%
  image_convert(type = "bilevel")

# get what the grayscale values are 
# for each pixel in the letter image
my_letter_pixels <- image_data(my_letter)[1,,] %>%
  as.numeric()

# go through each pixel in the letter image
# and randomly sample a photo that matches
# based on its shade
mosaic_photos <- c(1:length(my_letter_pixels))
for(i in 1:length(my_letter_pixels)){  
  gs_origin <- my_letter_pixels[i]
  target_photo <- photo_ref %>%
    filter(gs_shade == ifelse(gs_origin==0,"dark","light")) %>%
    sample_n(1) %>%
    pull(photo_num)
  mosaic_photos[i] <- target_photo
}

# make the mosaic!
mosaic_photos %>%
  image_grid()
```