* Standard Non-Standard Evaluation
Dplyr and other tidyverse tools are great for interactive data analysis, but can be difficult to program with. The problem stems from non-standard evaluation: this paradigm makes dplyr very expressive, but creates new class of problems for users to solve. This talk will look at standard non-standard evaluation techniques used to wrangle dplyr into user-defined functions.
* Gaining Client Trust in Predictive Models
Customers are excited by the prospect of using predictive models. But, when they are faced with trusting a model to make daily decisions they sometimes revert back to their old processes. At Avantas we produce thousands of models weekly, so we needed a solution that could solve this at scale. We will demonstrate how we worked with our clients and used R Markdown to deliver reporting that made sense, increased faith in the predictions, and decreased backlash to our internal support and account management teams.
* Open Forum
We will pose three questions to the group to start a discussion.
Do you think R is capable of being part of a production data pipeline?
○ What are people using for data visualization in R and for what purpose?
○ What did you learn first Tidyverse or Base R and how would you do it differently?
○ What is you favorite resource for learning R?
○ What databases do you use with R and how?