shiny-tutorial.Rmd

---
title: "Shiny app tutorial for PMed"
output: pdf_document
---

[`shiny`](https://github.com/rstudio/shiny) is a package for R to build interactive web apps, making sections of R code reactive. [Shiny](https://shiny.rstudio.com/) lets us use R to visualize data and incorportate those plots into a site.

# Getting started with R

1. Download an R distribution from [CRAN](https://cran.r-project.org/)

2. Download [Rstudio](https://www.rstudio.com/products/rstudio/download/#download)
    + explore [Rstudio](https://www.rstudio.com/online-learning/)

# R basics for today

## Packages

```{r eval=FALSE}
install.packages("shiny")
library(shiny)
```

## File paths: Working directory
```{r eval=FALSE}
getwd()
setwd([your path])
```

## Assignment
```{r}
x <- 5
x

x = 7
x
```

## Vectors
```{r}
x <- c(4,9,1,0)
x*3
x[1:2]

x <- c(4,"four")
x
```

## Lists
```{r}
x <- list(4,"four")
x
x <- list(value=4, word="four")

x$value
x[[1]]
x[["value"]]

x$french <- "quatre"
x$lower <- 1:3
x

x$lower <- seq(1,3.5,by=0.5)
x$isFour <- TRUE

x
```

# Example: shiny default
# Old Faithful geyser in Yellowstone National Park

* `File > New File > Shiny Web App ...`
    + `shiny::runApp('first-app')`
* Three functions:
    + ui()
    + server()
    + shinyApp()
* ui() define the inputs and displays the outputs
* server() defines the outputs with user-modified inputs
* shinyApp() takes both functions and excutes them, creating the web app
* Simple example using a built in dataset `faithful`
    + Waiting time between eruptions and the duration of the eruption for the Old Faithful geyser in Yellowstone National Park, Wyoming, USA.
    
Wait what are we plotting?
```{r}
?faithful
names(faithful)
head(faithful)
```

Lets make this shiny app more interesting ...

1. Change titles
    + change x value to be more readable ($waiting instead of [,2])
    + change plot title
    + remove "x" label
2. Add second histogram for second variable in dataset
3. Add second slider bar but make it a value input
4. Add titles to make it clear which is for which - p() vs h3()
    + add space with br()
5. Explore the relationship of time since last eruption and duration of eruption
    + plot regression line

```{r}
summary(lm(eruptions ~ waiting, data=faithful))
```


6. Oooh it looks like there are two groups.
    + add line to separate the two
7. Next steps ... do ML to find the best separatrix to maximize likelihood? minimize RMSE?

# Example: DDx Dashboard

Let's build the DDx app from scratch

# Example: PAWS

Data sets from [Philadelphia Animal Welfare Society (PAWS)](https://phillypaws.org/), a non-profit rescue organization providing vet care and adoption services for stray and surrendered animals.

RLadies repo forked at: [https://github.com/drscranto/2019_datathon](https://github.com/drscranto/2019_datathon)

# More to learn 

* Preset dashboards [`shinydashboard`](https://rstudio.github.io/shinydashboard/)
* Read data from an s3 bucket [`aws.s3`](https://github.com/cloudyr/aws.s3)
* Options for deploying a shiny web app 
    + [https://www.rstudio.com/products/shiny/shiny-server/](https://www.rstudio.com/products/shiny/shiny-server/)
    + Hosting, security, and scaling
    + Easy deployment with [`shinyapps.io`](https://www.shinyapps.io/)