Processing math: 100%

class: title_bg

.title[
Computation infrastructure for<br/>teaching Bayesian modeling
]

.conference[
  .name[ 
    JSM 2020 
  ]
  .bitly[ 
    [bit.ly/JSM2020_Sta444](https://bit.ly/JSM2020_Sta444)
  ] 
]

.author[
.name[
Colin Rundel
]
.school[
Univ of Edinburgh<br/>
Duke University
]
]

---

## Context

Sta 444 / 644 - Spatiotemporal Modeling

* 4th year undergraduate elective (2nd year MS elective)

* *Prereq* Sta 360 - Bayesian Inference and Modern Statistical Methods

* Weekly labs / problem sessions with TA

* 5 hws + 1 group project over the semester <br/>
(mathematical, computational, and applied problems)

---

## Learning Outcomes

* Modeling methods (lm, glm, ARIMA, GPs, CAR, etc.)

* Model assessment and validation

* Bayesian model implementation (probabilistic programming)

---

.columns[
.col[
DLM for predicting `\(\text{CO}_2\)`

<img src="imgs/dlm-1.png" width="80%" /><img src="imgs/dlm-2.png" width="80%" />
]

.col[
Violent Crime in Baltimore City

<img src="imgs/baltimore1.png" width="57%" style="display: block; margin: auto;" /><img src="imgs/baltimore2.png" width="57%" style="display: block; margin: auto;" />
]
]

---

## Software

* Wide varierty of choices: BUGS, Jags, Stan, etc.

* Some considerations:

* Generalizability vs. specificity
  
  * Syntactic complexity
  
  * Performance
  
  * Limitations

---

## Computational Complexity and Efficiency

This is often the first course where students engaged with models that cannot be fit "instantly".

* Basics of algorithmic complexity

* Model limitations vs software limitations

* Implementation vs run time

---

## Infrastructure

The platform provided for students matter (labs vs. servers vs. student laptops):

* Configuration / Administration

* Performance

* Workflows

.footnote[
Çetinkaya-Rundel & Rundel (2018) - [Infrastructure and Tools for Teaching Computing Throughout the Statistical Curriculum](https://amstat.tandfonline.com/doi/abs/10.1080/00031305.2017.1397549)
]

---

# Thank you!

.middle[ .center[

<div style="width: 98%">
<table class="contact" style="text-align: left; margin-left:auto; margin-right:auto; width:50%;">
<tbody>
<tr><td><br/></td><td>&nbsp;&nbsp;&nbsp;</td></tr>
<tr>
  <td style="vertical-align: middle;"> <i class="fas fa-envelope fa-fw fa-2x"></i> </td>
  <td></td>
  <td> 
    <a href="mailto:rundel@gmail.com">rundel@gmail.com</a>
  </td>
</tr>
<tr><td><br/></td></tr>
<tr>
  <td style="vertical-align: middle;"> <i class="fab fa-twitter-square fa-fw fa-2x"></i> </td>
  <td></td>
  <td> 
    <a href="https://twitter.com/rundel">@rundel</a>
  </td>
</tr>
<tr><td><br/></td></tr>
<tr>
  <td style="vertical-align: middle;"> <i class="far fa-file-powerpoint fa-fw fa-2x"></i> </td>
  <td></td>
  <td> 
      <a href="http://bit.ly/JSM2020_Sta444">bit.ly/JSM2020_Sta444</a>
  </td>
</tr>
<tr><td><br/></td></tr>
<tr>
  <td style="vertical-align: middle;"> <i class="fas fa-chalkboard-teacher fa-fw fa-2x"></i> </td>
  <td></td>
  <td> 
    <a href="http://www2.stat.duke.edu/~cr173/Sta444_Fa18/">stat.duke.edu/~cr173/Sta444_Fa18/</a>
  </td>
</tr>
</tbody>
</table>
</div>
] ]

Notes for current slide

Notes for next slide

Computation infrastructure for
teaching Bayesian modeling

JSM 2020

bit.ly/JSM2020_Sta444

Colin Rundel

Univ of Edinburgh
Duke University

1 / 8

Context

Sta 444 / 644 - Spatiotemporal Modeling

4th year undergraduate elective (2nd year MS elective)
Prereq Sta 360 - Bayesian Inference and Modern Statistical Methods
Weekly labs / problem sessions with TA
5 hws + 1 group project over the semester
(mathematical, computational, and applied problems)

2 / 8

Learning Outcomes

Modeling methods (lm, glm, ARIMA, GPs, CAR, etc.)
Model assessment and validation
Bayesian model implementation (probabilistic programming)

3 / 8

DLM for predicting $\text{CO}_2$

Violent Crime in Baltimore City

4 / 8

Software

Wide varierty of choices: BUGS, Jags, Stan, etc.
Some considerations:
- Generalizability vs. specificity
- Syntactic complexity
- Performance
- Limitations

5 / 8

Computational Complexity and Efficiency

This is often the first course where students engaged with models that cannot be fit "instantly".

Basics of algorithmic complexity
Model limitations vs software limitations
Implementation vs run time

6 / 8

Infrastructure

The platform provided for students matter (labs vs. servers vs. student laptops):

Configuration / Administration
Performance
Workflows

Çetinkaya-Rundel & Rundel (2018) - Infrastructure and Tools for Teaching Computing Throughout the Statistical Curriculum

7 / 8

Thank you!


		rundel@gmail.com

		@rundel

		bit.ly/JSM2020_Sta444

		stat.duke.edu/~cr173/Sta444_Fa18/

8 / 8

Context

Sta 444 / 644 - Spatiotemporal Modeling

4th year undergraduate elective (2nd year MS elective)
Prereq Sta 360 - Bayesian Inference and Modern Statistical Methods
Weekly labs / problem sessions with TA
5 hws + 1 group project over the semester
(mathematical, computational, and applied problems)

2 / 8

Paused

Help

Keyboard shortcuts

↑, ←, Pg Up, k	Go to previous slide
↓, →, Pg Dn, Space, j	Go to next slide
Home	Go to first slide
End	Go to last slide
Number + Return	Go to specific slide
b / m / f	Toggle blackout / mirrored / fullscreen mode
c	Clone slideshow
p	Toggle presenter mode
t	Restart the presentation timer
?, h	Toggle this help

Esc	Back to slideshow