Data Science - Delivered Continuously - XConf 2017
-
Upload
christian-deger -
Category
Data & Analytics
-
view
135 -
download
3
Transcript of Data Science - Delivered Continuously - XConf 2017
![Page 1: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/1.jpg)
Arif Wider & Christian Deger
DATA SCIENCE,DELIVERED CONTINUOUSLY
![Page 2: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/2.jpg)
A CONFERENCE ALL ABOUT TECHNOLOGY
![Page 3: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/3.jpg)
Christian DegerChief [email protected]@cdeger
![Page 4: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/4.jpg)
Arif WiderSenior Consultant/[email protected]@arifwider
![Page 5: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/5.jpg)
PL
S
RUS
UA
RO
CZ
D
NL
B
FA
HRI
E
BG
TR
18countries
2.4m+cars & motos
10m+users per
month
![Page 6: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/6.jpg)
The task: A consumer-facing data product
6XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 7: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/7.jpg)
The task: A consumer-facing data product
7XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 8: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/8.jpg)
The task: A consumer-facing data product
8XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 9: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/9.jpg)
The prediction model: Random forest
9
Volkswagen GolfCar listings oflast two years
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 10: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/10.jpg)
How to turn an R-based prediction model into a high-performance web application?
10
?
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 11: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/11.jpg)
How to turn an R-based prediction model into a high-performance web application?
11XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 12: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/12.jpg)
How to turn an R-based prediction model into a high-performance web application?
12XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 13: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/13.jpg)
How to turn an R-based prediction model into a high-performance web application?
13
Continuous Delivery!
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 14: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/14.jpg)
Application code in one repository per
service.
CI
Deployment packageas artifact.
CD
Deliver package to servers
Typical delivery pipeline
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 15: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/15.jpg)
Continuous delivery pipelines
15
Prediction Model Pipeline
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 16: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/16.jpg)
Continuous delivery pipelines
16
Prediction Model Pipeline
Web Application Pipeline
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 17: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/17.jpg)
The price for CD: Extensive model validation
17XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 18: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/18.jpg)
The price for CD: Extensive model validation
18XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 19: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/19.jpg)
Lessons learned
19
Form a cross-functional team of data scientists & software engineers!
Software engineers… learn how data scientists work… and understand the quirks of a prediction model
Data Scientist… learn about unit testing, stable interfaces, git, etc.... get quick feedback about the impact of their work
Model and product iterations become much faster!
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 20: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/20.jpg)
Lessons learned
20
Generating gigabytes of Java code
is a challenge for the JVM
Use the G1 garbage collector
Turn off Tiered Compilation
Do extensive warm-ups
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 21: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/21.jpg)
Lessons learned – Warm up
21XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 22: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/22.jpg)
Lessons learned
22
The approach of applying Continuous Delivery to
Data Science is useful independently of the tech
Successfully applied similarly to a Python- and
Spark-based project
Even more useful when quick model evolution
is required because of rapidly changing inputs
(e.g. user interaction)
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 23: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/23.jpg)
Conclusions
23
Continuous Delivery allows us to bring prediction
model changes live very quickly.
Only extensive automated end-to-end tests
provide confidence to deploy to production
automatically.
Java code generation allows for very low response
times and excellent scalability for high loads but
requires plenty of memory.
XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 24: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/24.jpg)
Conclusions: Price evaluation everywhere
24XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 25: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/25.jpg)
Conclusions: Price evaluation everywhere
25XConf ’17 Hamburg/Manchester Data Science, Delivered Continuously – A. Wider & C. Deger
![Page 26: Data Science - Delivered Continuously - XConf 2017](https://reader031.fdocuments.net/reader031/viewer/2022030317/5a65b3c67f8b9a0a5f8b499b/html5/thumbnails/26.jpg)
QUESTIONS?
THANK YOU
Arif Wider & Christian Deger
@arifwider @cdeger