Data Scavenger Hunts

class: center, middle, inverse, title-slide

# Data Scavenger Hunts
## Learning about datasets together
### Ted Laderas
### 2019-05-09

---

# Overview: Data Scavenger Hunts

- Introduction
- Why?
- What are they?
- Who are they For?
- Who are they for? (reprise)
- Call to Action

---
# Who am I?

![Who am I?](image/giphy.gif)

- Assistant Professor at Oregon Health & Science University
- Many years working with bioinformatics data
- I want people to understand the data they generate
- I love teaching visualization

---
# Why? Looking at Data Together

- Visualization is empowering

--
- Visualization is accessible

--
- Visualization lets us have a conversation

--
- Visualization/Interpreting is a valuable skill

---

# What are Data Scavenger Hunts?

.pull-left[
- Social learning activity about a dataset
- Inspired by Exploratory Data Analysis
  - 'Find patterns, reveal structure (Behrens)'
- Understand associations between variables
  - How is Body Mass Index associated with Diabetes?
- Reflect/present discoveries to the group
]

---
# What? Burro

.pull-left[
- Burro[w] into the data!
- R Package/Shiny App for exploring many kinds of data
- Spin up a web app for any dataset
- Maps variable types to the appropriate visualizations
- Allow people to explore the data quickly
]

.pull-right[
![Jessica Minnier and Ted Laderas](image/burro3.png)
]

.footnote[
Jessica Minnier and Ted Laderas
]

---
# What? Burro Overview

.pull-left[
- Organized by variable types
- Overview Data
- Continuous vs Categorical Data
- Assess single variables first, then look at interactions by variable type
]

.pull-right[

<img src="image/vis_framework.png" width="267" />
]

---
# What? Burro Example

Looking for association of episodes of little interest with depressive episodes:

![](image/categorical.gif)

---
# What? The Questions

- Have pre-prepared questions that guide exploration
- Questions range from simple to complex
- Map question types into visualization types

---
# What? Try it out

http://bit.ly/csv_burro

Try one of these questions:

- Why is age capped at 80 years in this dataset?
- Is marijuana use associated with depressive episodes?
- Are hours of sleep associated with depressive episodes?

---
# What: Reflection/Discussion

- Bring back our observations to the group
- Simple Template:
  - What was our question?
  - What was our evidence? How did we find it?
  - What other variables should we look at?
- Group reflection: what did we learn as a group?

---
# Who has participated?

- Graduate Students
  - Portland State Students
  - Cardiovascular Risk Data (synthetic data sets)
- Graduate Students (clinical data wrangling course)
  - Clinical and Bioinformatics incoming students
  - Sleep Heart Health Study Data
- BioData Club Members (NHANES Scavenger Hunt)
  - NHANES Data
- Undergraduate Students (Data Literacy)
  - Public Health Education
  - NHANES Data
  
---
# What do they say (BioData Club)?

- "The scavenger hunt. It is great place to apply this knowledge."
- "I enjoyed the collaborative and self-paced approach."
- "I liked the dataset and the open curiosity element of the workshop."

---
# Who: Useful for Citizen Data Efforts?

- What would it take to make `burro` useful for citizen data efforts?
    - Better "just in time" documentation
- Can it be used to teach graph literacy and data curiosity?
- Can we use this framework to democratize data science?

---
# Who: Spatial Data Exploration

- New example: Social Determinants of Health
- Prototyping with spatial data frames (shapefiles, sf)
  - Example: https://tladeras.shinyapps.io/oregon_county/
---

# Call to action

We need beginners/novices!

- Need help with usability testing/documentation
- Need help with workflows for different data types
- Need help with standardizing template for scavenger hunts

---
# Stay in Touch

These slides: http://bit.ly/csv_hunt

- email me: laderast@ohsu.edu 
- @tladeras
- More about data scavenger hunts and `burro`:
    - http://laderast.github.io/burro/
- Burro email list: 
    - http://tinyletter.com/tladeras
- Contributing Info:
    - https://laderast.github.io/burro/CONTRIBUTING
- BioData Club: 
    - Local portland club that focuses on learning about data
    - http://biodata-club.github.io