class: left, top, title-slide .title[ # Presenting Predictive Analytics Projects ] .author[ ### Ken Arnold
Calvin University ] --- class: center, middle ## Making a Data-Driven Argument --- ## Key points - Consider the *audience* to get the level of detail right. - Never assume your audience can rapidly process complex visuals. ([Claus Wilke](https://clauswilke.com/dataviz/telling-a-story.html#make-a-figure-for-the-generals)) - Consider the *purpose* to choose report vs dashboard vs presentation - Anchor *claims* in *data*. - Tell stories - "but-therefore" - Rational Reconstruction --- ### Make a point Which is clearer? .pull-left[ **Report A** MAIN POINT * Supporting chart 1 * Supporting table 2 * Supporting model 3 Discussion about how each supports main point ] .pull-right[ or **Report B** * Chart 1 * Table 2 * Model 3 * Chart 4 * Table 5 * Chart 6 * Model 7 * Chart 12 * Table 25 ] *Simplicity, clarity, depth of understanding*, not *exhaustiveness* --- ### Tell a Story - *Chart* ... - **Therefore**, *approach* ... - **but** *chart*, *table* - **so** ... [but-therefore](https://www.youtube.com/watch?v=vGUNqq3jVLg) See also: "[Telling a story and making a point](https://clauswilke.com/dataviz/telling-a-story.html)" --- ### ... but not the *history* - Tell the "**[rational reconstruction](https://web.stanford.edu/class/cs224u/readings/shieber-writing.pdf)**" of the story of how the analysis was done. - Make it seem like your approach was just simple common sense - ... even if you actually had to try out a lot of different things on the way to getting there. - Start with a **simple approach** - Analyze its successes and failures - Increase complexity thoughtfully and deliberately. - Compare simple with complex - **Don't** include every model you tried, every idea you had... - **Do** report on a few (carefully selected) --- ### Anchor conclusions in data .pull-left[ * The units are probably seconds<br><br> * The fit looks good<br><br> * This was surprising ] -- .pull-right[ * because the median, 600, would be 10 minutes * because the mean error of $15 is less than 0.1% of the price * because I expected that people would leave higher ratings on products they enjoyed more ] --- ### Use appropriate language .pull-left[ **Plain language** for the overview, conclusion, and visuals. * Labels in visuals: use real names, not `code_names`. (For all aesthetics, not just x and y.) * Don't assume the reader knows the structure of the data. ] .pull-right[ **Technical language** when describing methods (data acquisition, wrangling, modeling, etc.). * What data representation choices did you make? *why*? * What modeling choices? Why? etc. ] --- ### Some color tips <https://blog.datawrapper.de/beautifulcolors/> --- class: center, middle ## Start Simple! --- ## Tools for Communication - Markdown: **know** your *formatting*, including `- lists` and `[links](URL)`. - Data graphics: `ggplot`, `plotly`, [`rbokeh`](https://hafen.github.io/rbokeh/articles/rbokeh.html) - Slides: - These slides are `xaringan` + [`xaringanthemer`](https://pkg.garrickadenbuie.com/xaringanthemer/index.html) + [`xaringanExtra`](https://github.com/gadenbuie/xaringanExtra/) - [Other options](https://bookdown.org/yihui/rmarkdown/presentations.html) include `ioslides`, `slidy`, ... - but PowerPoint / Google Slides work fine too. - Getting on the web - GitHub Pages - `flexdashboard`: <https://pkgs.rstudio.com/flexdashboard/articles/examples.html> - Shiny apps - RStudio Connect --- ### Example: [Shiny](https://shiny.rstudio.com/) Apps <https://shiny.rstudio.com/gallery/> Further reading: [Engineering Production-Grade Shiny Apps](https://engineering-shiny.org/) --- ### Report details - Eschew extraneous verbiage - Cite sources clearly - Data - Existing analysis you're critiquing - Any nontrivial code you've used - Check the rendered document - common mistake: forgetting blank line after code chunk (messes up headers etc.) - figures legible - all elements included? - code legible (doesn't run off the page) - ...but understandable without reading the code!