README.Rmd

---
title: "dfvis"
output: md_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, fig.width = 12, fig.height = 8)
```

**WIP: IN DEVELOPMENT**

#### A ggplot2 based implementation of tabplot ([github repo](https://github.com/mtennekes/tabplot), [paper](http://www.jds-online.com/file_download/379/JDS-1108.pdf))

tabplot offers a fast way to eyeball dataframes (my go-to tool over years). This uncovers possible interactions between variables when sorted by some variable. Hence, it builds intuition for any further modeling.

## What is different from tabplot

- Adds out-of-box support for grouped tibbles (tidy dataframes)
- Based on ggplot for flexible geoms for different variable types
- dfvis might not be as fast as tabplot

## Illustrations

```{r}
pacman::p_load("dplyr", "tabplot", "dfvis")
```

```{r}
data("attrition", package = "modeldata")
attrition = as_tibble(attrition)
attrition_6 = attrition[, 1:6]
skimr::skim(attrition_6)
```

### Ungrouped case

```{r}
autoplot(attrition_6, sort_column_name = "DistanceFromHome")
```

### Grouped Case

```{r}
suppressWarnings(
  attrition_6 %>% 
    group_by(Attrition) %>% 
    autoplot(sort_column_name = "DistanceFromHome")
  )
```

### tabplot

```{r}
tabplot::tableplot(attrition_6, sortCol = "DistanceFromHome", nBins = 10)
```

## Development and Contribution

- Contributions are welcome!
- Create interactive version (with shiny?)