Skip to content

Latest commit

 

History

History
199 lines (159 loc) · 5.3 KB

plan_bandung.md

File metadata and controls

199 lines (159 loc) · 5.3 KB
Raw

Event

Gentle Introduction to Developing modern web apps in R & Python WiFi - Block 71 Bandung - Username: blk71 - Password: blk71bdg

- Username: blk71daily
- Password: innovationfactory

Schedule and Deliverables: https://algorit.ma/kickstart-introduction-developing-modern-web-apps/

Web App template (download link): https://github.com/onlyphantom/darkershiny

Objective

  • What do we want in a "modern" web app?

    • Functionalities
      • Work with any kind of input sources
      • Get machine learning functionalities
        • Prediction
      • Get visualization functionalities
        • Embed visualization into our app
      • Security features (production code)
      • Cloud deployment
      • Version control
      • Authentication
      • [Optional] Administration
    • Appearance and UX
      • Mobile-friendly
      • Customized with own CSS / JS

  • How do we gain the skills required for objective #1

Plan

Learning Resources

Demo 1:

  • TensorFlow as a web app (HousingApp)

Demo 2:

  • Shiny on SQL (Limitless)

Demo 3:

  • Shiny in action (Quadrant)

Demo 4:

  • Authentication and Continuous Delivery (Pedagogy)

Demo 5:

  • Telco app (Confidence)

Building a web app in Shiny (Start to Finish)

  1. Download the web app template

  2. Unzip the file

  3. Right-click on app.R and open > open in RStudio

  4. Click on "run app" green button in RStudio

  5. Click on "open in web browser"

  6. Open RStudio

  7. Click on New > New File > Shiny Web App

  8. Click on "Run App"

  9. Create our Visualization plot(x, y) or ggplot()

  10. Assign that plot to a variable myplot <- plot(x,y)

  11. Use renderPlot({ myplot }) to render plot in Shiny

  12. Click on the Publish button next to Run app

Building a web app in Python (Start to Finish)

  1. Use a microframework like Flask
    • If you don't have it installed yet: pip install flask
  2. Use virtual environments
    • Or alternatively: conda environments
  3. Use a modern IDE
    • Has github integration
    • Simple code completion / intellisense
  4. Create app.py and the boilerplate:
from flask import Flask
app = Flask(__name__)

@app.route('login')
...

Use flask run to run the app

Difference between creating web apps in R vs Python

  • Python: In a microframework or framework like Django or Flask, you write separate code for each of:

    • html
    • js
    • css
    • py

  • R: Use shiny and write everything in R

    • app.R

  • In R: two options for deployment

    • Free shinyapps.io (Publish button)
      • samuelc.shinyapps.io/predictor
    • Shiny Server (open source) deployed on an Ubuntu machine

  • In Python:

    • pythonanywhere
    • Docker (container): alpine linux
    • Run on Linux machine

  • Advantages

    • R

      • data manipulation if often one line of code and built in without any reliance on any external libraries
      • ggplot2 and other cool visualization libraries
      • Shiny is super easy to use (literally three steps)

    • Python

      • Ease of deployment (any linux server or container)
      • Access to pandas, numpy, sk-learn, tf..
      • More "general" so a lot of libraries outside of the data science world

  • R and Python are both equally good choice

    • Any big companies releasing open source AI tools
      • Have both R and Python

Concrete Steps

  1. Pick up visualization

    • R: lattice plot, ggplot and base plot in R
    • Python: matplotlib

  2. Pick up machine learning

    • Sk-learn, tensorflow or keras or mxnet, caret or algorithms implemented in R

  3. Get familiar with Shiny or Flask

  4. Learn cloud deployment and version control

    • Choices:

      • AWS (Amazon web services)
      • Microsoft Azure
      • Google Cloud
      • ...

    • Sign up for github.com

  5. Reference Materials