Why bother?
Why should you DLAB?
Pretty!
Organized
Interactive
Reproducible
Free
You get to use R
Flexible
Better visualizations (sorry gg)
People will respect you
How to DLAB?
A 12 Step Program to Quit Emailing Plots & Csvs
How to DLAB?
A 12 Step Program to Quit Emailing Plots & Csvs
- Basic Layout
- Printing: Tables,Descriptice Stats, Model summaries
- Interactive Datasets
- Interactive Plots
- Exporting
- CSS basics
- Branding (Logos, font, colors)
- Header/Footer
- Hosting
- Automate-ish (render + system())
- Replicate (render with arguments)
- Automate or App it (cron, shiny runtime)
How to DLAB?
A 12 Step Program to Quit Emailing Plots & Csvs
1. Basic Layout
2. Printing: Tables, Descriptice Stats, Model summaries
3. Interactive Datasets
4. Interactive Plots
5. Exporting
1. Basic Layout
<3 flexdashboard <3
.Rmd + flexdashboard
Before :(
After :)
flexdashboard layouts
My go-to
optional:
vertical_layout: {scroll, fill}
orientation: {columns, rows}
print() is simply unacceptable
2. Printing Tables, Descriptive Stats & Model Summaries
Tables, Descriptive Stats & Model Summaries
printr summarytools::dfSummary stargazer::stargazer
Table Printing: my go-to
tables & descriptive stats: printr
model summaries: stargazer::stargazer
Model Summary & Comparisons
stargazer::stargazer
Which model summary statistics do you wantto include in the output? Bookmark the link for later
https://rdrr.io/cran/stargazer/man/stargazer_stat_code_list.html
3. Interactive Datasets
Interactive Datasets
DT::datatable
Datasets
DT::datatable
pretty printing
flexible formatting
column filtering
csv download
cell highlighting (think heatmap)
4. Interactive plots
We're not in PowerPoint anymore.
4. Interactive plots
highcharter
ggplot2 will always have a place in your heart, but dashboards need more than pngs
4. Interactive plots
highcharter
Other options
- Maps: leaflet
- Timeseries: dygraphs
- ggplot2 compatible: plotly
4. Exporting
Exporting
Datasets with DT
Highcharter Plots
Dashboard LIKE A BOSS: pro tips
Let your dashboard work for you
SIMPLE
- Start with a simple template & someone else's css
- Add in the essentials or the lowest hanging fruit as you go
ORGANIZED
- Maintain a folder containing: css, exporting scripts, color palletes, global themes, & templates
- Keep a staging folder for exploration–don't make changes you won't know how to undo.
- Keep expensive data cleaning in a separate script (easier to introduce it later than remove it after)
EXPLORE
- Try something new (template, plots, or css) to improve the appearance & usability of each new dashboard
- Slowly work in: About page & explanations, interactive viz, exporting, logo, footer, padding/margins
Resources
github.com/maloriejhughes
- Basic Layout:
- R Markdown + flexdashboard
- Interactive Viz:
- All: highcharter
- Maps: leaflet, my leaflet tutorial
- Timeseries: dygraphs
- ggplot2 compatible:plotly
- Data/Tables
- Datasets: DT::datatable
- Tables & Summaries: printr
- Meh IMO: summarytable
- Models
- Model Summaries: stargazer
- Exporting Data & Plots
- highcharter
- custom script
- CSS (global settings: themes, layout, background, etc)
- html ( in-line style/text changes, images, hyperlinks, web basics )
- Branding ( css+html: custom fonts, global color themes, watermarks, etc)
- Hosting
- Automate-ish (render + system())
- Replicate (render with arguments)
- Automate or App it (cron, shiny runtime)