RaukR presentation - vikingr demo

class: center, middle, inverse, title-slide

# RaukR presentation - vikingr demo
## RaukR 2019 • Advanced R for Bioinformatics
### Ben Pontiller, Andreas Schneider, Markus Skyttner

---

exclude: true
count: false

---
name: intro
class: middle, spaced

## Introduction

### "Decode this AIS message"

> .large[1560376557, !AIVDM,1,1,,A,13uDcBP0001CcmjPvca;d6Cn0<09,0*27,1121886915]

### Reference: [AIS format - https://gpsd.gitlab.io/gpsd/AIVDM.html](https://gpsd.gitlab.io/gpsd/AIVDM.html)

> .large[More than 50 lookup tables and descriptions of possibly valid approaches for parsing along with some sample data]

---
name: intro
class: middle, spaced

## Introduction

### AIS: Automatic Identification System

+ intended to assist a vessel's watchstanding officers and allow maritime authorities to track and monitor vessel movements
+ info such as unique identification, position, course, and speed; sent in regular intervals by every vessel

---
name: inspiration
class: middle, spaced

## Inspiration

* One or many messages - baked into lines in a log file - are to be read and decoded 
* Tidyverse `readr` offers `read_lines()` and can read from text streams
* If there are parsing issues - inspect them with `readr::problems()`

> Let's do it like that!

* The reference materials mentions `ais.py` - a file somewhere on the Internet
* Let's use that one - maybe we can call it with `reticulate` from R?

---
name: lessons
class: middle, spaced

## Lessons

### Data and python issues

* Laptop crash! -> new laptop - this old one doesn't like `reticulate` - memory leaks?
* Log example data contains (timestamp, message) tuples but also other junk
* Big differences in data between ships (7000/10000 from one little coast guard boat, NAs)
* Wrapping `ais.py` needs a timestamp to link messages to the original log
* So this needs a small edit of the `ais.py` file to add this to the JSON output
* Turns out `ais.py` works well but the JSON could be more detailed

### Package development

* Rebuilding vignettes and reloading package documentation is shaky
* Live with restarting R session and wiping environment for the vignettes
* A vignette can be a flexdashboard - but can get big embedding data, would annoy CRAN

---
name: results
class: middle, spaced

## Results

Repo: [https://github.com/mskyttner/vikingr](https://github.com/mskyttner/vikingr)
Cloud-friendly distro: [rstudio.cloud + vikingr etc](https://hub.docker.com/r/mskyttner/raukr/tags)

### Functions

* Function to read the log
* Function to decode the AIS message(s)

### Data

* Sample data from Visby Harbor
* Should work with other sample dataset

### Visuals

* Flexdashboard with leaflet map

> .large[Try it from [https://mskyttner.github.io/vikingr/vikingr-x.html](https://mskyttner.github.io/vikingr/vikingr-x.html)!]

Bonus: [Dockerized](https://github.com/mskyttner/vikingr/tree/master/docker)

---
name: thanks
class: middle, spaced

# Thank you. Questions?

R version 3.4.4 (2018-03-15) Platform: x86_64-pc-linux-gnu (64-bit)OS: Ubuntu 18.04.2 LTS

Built on : 20-jun-2019 at 09:05:54

__2019__ • [SciLifeLab](https://www.scilifelab.se/) • [NBIS](https://nbis.se/)