dynamic.Rmd

---
title: 'Iris: Sepal-Ratio Modeling'
author: "Bastiaan Quast"
date: "Thursday, July 18, 2014"
output:
  ioslides_presentation:
    keep_md: yes
    transition: faster
runtime: shiny
---


# We want to model sepal ratios, do we need separate models for each species?


## Iris dataset

- Included in R package **datasets**
- Originally by Fisher or Anderson
- Measurements for **petal** and **sepal** width and length
- Fifty flowers of three species
    - Setosa
    - Versicolor
    - Virginica

```{r}
str(iris)
```


## LOESS Model

 - **Lo**cally W**e**ighted **S**catterplot **S**moothing
    - [more info on LOESS](http://en.wikipedia.org/wiki/Local_regression)
 - General model and species-specific models
    - Standard Error bands
    - Grey band for general model
    - Coloured bands for respective species
 - Adjustable 'wobblyness' of models
 - Adjustable point size and opacity


## Results
```{r echo=FALSE, message=FALSE}
# load the ggvis package
library(ggvis)

# load the dplyr package
# for use of magrittr pipe: %>%
library(dplyr)

# lazyload the dataset and pass it to
iris %>%
  # the ggvis function, now set the axis
  ggvis(~Sepal.Width, ~Sepal.Length) %>%
  # add the smoother for all points
  layer_smooths(span = input_slider(0.5, 1, 0.5,
                                    label = 'General Model'),
                se = TRUE) %>%
  # now group by species
  group_by(Species) %>%
  # add points per species and colour them
  layer_points(size := input_slider(10, 200, 50,
                                    label = 'Point Size'),
               opacity := input_slider(0.5, 1, 0.5,
                                       label = 'Point Opacity'),
               fill = ~Species) %>%
  # now add smoothers per species and colour them
  layer_smooths(span = input_slider(0.5, 1, 0.5,
                                    label = 'Species Models'),
                fill = ~Species, se=TRUE) %>%
  # add new labels
  add_axis("x", title = "Sepal Width") %>%
  add_axis("y", title = "Sepal Length") %>%
  # finally declare as shiny object
  bind_shiny("ggvis", "ggvis_ui")
```