R Shiny is free, but is it the right tool for your needs? With this talk you will learn to estimate your needs and the constraints, often underestimated, of using an open source solution.
1. The Right Tool?!
R Shiny!
Xavier Prudent – Data Scientist!
!
Data Montréal Meetup, 12th June 2018!
2. Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
To Do or Not To Do?!
Pictures : heineken
3. R !
hundreds of packages, !
advanced statistics, visualization,!
machine learning…!
!
Web interface!
!
Interactive analysis (click & see)!
!
Highly modular!
!
Keep full control on data, performance…!
!
Easy to use!
!
FREE!
R Shiny : an Awesome Tool to Share Analysis!
Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
4. I want to speed up my analysis work!
!
I want to concentrate more on the analysis than the coding!
!
!
I want an in-house tool for my collaborators!
!
!
My boss wants to go through the data by himself!
!
!
My client wants to go through the data by himself!
!
!
!
!
!
!
!
Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
1. State Your Needs!
What about Excel / Google Doc ?!
5. Input!
My data files are large!
My data is changing regularly!
Your choice should be driven by your strategy more than your budget
Hiring a programmer / Data Scien:st may not be cheaper
2. State Your Constraints!
UX!
It must look professional!
It must be user-friendly!
Access point!
It must be available online!
It must be secure !
Method!
I need a special visualization /
statistical method!
Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
8. Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
3. What you need: Time!
Medium analysis with clean data : 40h!
(no machine learning)!
!
If large data and need to speed up, !
If need to tune the design : + 40h!
Something to show to a client
and not be ashamed of : !
80h!
Hired programmer : 30$/h!
Freelance programmer : 60$/h!
3 weeks!
2k – 5k$!
9. 3. What you need: Overview!
Material : 2 weeks!
1k – 2k$!
Implementation: 3 weeks!
2k – 5k$!
Total:!
2 months !
3k – 8k$!
10. R!
!
R Shiny!
!
Css, javascrip,t, html!
!
Linux, servers!
UX!
!
Data Science!
!
Package building!
!
Collaboration!
Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
3. What you need: Skills!
11. Accessing your data : think global!!
Firewall, servers, ports : the devil lies in the details!
Design : make a quick testing version!
Modularity : splitting your code is necessary AND painful!
Monitoring : R Shiny will keep it shy!
Balance your wishes : flexible does not mean easy!
Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
Frustration to be prepared for!
12. !
Do not do EVERYTHING by yourself!
Get helped by a network admin, a UX designer!
!
Use Plotly graphs for full interactivity!
!
!
Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
Quick & cheap also possible!
!
Make it a strategy!
!
Who is going to manage this project&!
!
Who is going to maintain this project?!
!
48 Junior Rule!
Give me a time estimate…!
Have you already done it?!
Yes -> x 4 !
No -> x 8!
!
A few tips!
Start simple
13. If total score 20 – 30 : consider R Shiny!
Xavier Prudent – Civilia – Data Montréal Meetup – 12th June 2018!
Should you go for R Shiny?!
Question! Score +3! Score +2! Score +1!
User?! You! Collaborator! Client!
Frequency of use! Day! Week! Month!
Needed flexibility! High! Middle! Low!
Needed Specialization! High! Middle! Low!
Place of this project in
your business model!
Long term! Middle term! Short term!
Data: size and
complexity!
Large! Middle! Small!
Project deadline! Long! Middle! Short!
Computing skills! High! MIddle! Low!
Material available! Server! Local network! Laptop!
In-house person for
maintenance!
Yes! Maybe! No!