Page Comparison

Streamlit (https://streamlit.io/)

Python

• Open source

• Free

• No

Yes…? (https://blog.streamlit.io/how-to-build-a-real-time-live-dashboard-with-streamlit/ )

• Beginner-friendly

• Simple API with excellent documentation

• Can mix visualizations into one presentation

• Can re-use Python code instead of rewriting it

• Requires a bit of HTML knowledge to truly customize

• Reruns the entire app once the input is changed

• Uses a localhost for local development

• Streamlit Cloud allows you to run applications in the cloud, share with colleagues

• https://streamlit.io/security

• You can display images, audio, and videos with Streamlit (Python Tutorial: Streamlit | DataCamp)

• You can input widgets

• Display progress and status bars

• Can show data visualization and maps

• Can make machine learning applications

Dash by Plotly (Dash Documentation & User Guide | Plotly)

Python

• Open source, MIT licensed

• Free

• Dash Enterprise (ie. deployment server) has an associated cost → have to fill out some form on the website to learn more about pricing

• Possible to run without internet connection, but all resources need to be bundled within the app

Yes (Live Updates | Dash for Python Documentation | Plotly)

• Pure Python → No Javascript unless you want to include Javascript assets

• Jupyter and Django integrations

• Cross-filtering/interacting with Plotly charts

• Aesthetics are more flexible, compatible with Bootstrap

• Has “callback” functions

• Robust modules library

• Difficulty with interactivity

• Difficulty to customize API

• No HTML → Must use Markdown function

• Cannot have two Python callbacks update the same element

• Limited colour options

• Plots that are shared are visible to everyone

• Takes time to load, a little slow

• Dash App Manager can be installed in cloud (Azure, AWS, GCP) (Dash App Manager (plotly.com))

• Can connect Dash App to SQL Database (Connect a Dash App to an SQL Database | Dash for Python Documentation | Plotly)

• There’s an app manager that’s available with Dash Enterprise (read more: Dash App Manager (plotly.com))

Seaborn (https://seaborn.pydata.org/index.html )

Python

• Open source

• Free

• No

• Seaborn has datasets that are available on Github → Require internet to load the datasets into Seaborn

• Not really → Uses matplotlib, so you need a while loop to use “real time data”

• Can install other libraries while using Seaborn since Seaborn is based on matplotlib

• Generates engaging plot to represent our data

• Feed our data using replot() method, library computes the values and places them without us worrying it

• Default themes are aesthetically pleasing

• Visualizes information in matrices and DataFrames

• Estimators are not suited for small datasets

  • Calculating and plotting confidence intervals uses bootstraps

  • Small datasets have inaccurate intervals since bootstraps are suited for large datasets

  • Have to calculate the intervals ourselves

• May need to reformat data (ie. Three common seaborn difficulties | by Michael Waskom | Medium) since Seaborn does not plot datasets normally

• Two different plot functions and difficulty plotting data other than categorical

  • Overview of seaborn plotting functions — seaborn 0.12.2 documentation (pydata.org)

• Based on matplotlib, no connection

• Based on matplotlib

• Close integration with pandas

• Dataset oriented API for examining relationships between multiple variables

• Specialized support for categorical variables to show observations or aggregate statistics

• Concise control over matplotlib figure styling with several built-in themes.

• Tools for choosing color palettes to reveal data patterns

Bokeh (https://bokeh.org/ )

Python

• Open source

• Free

• No..?

• Server introduction — Bokeh 3.0.3 Documentation

• Yes, supports streaming and real-time data

• Plots are flexible, interactive, and shareable

• Very Python centric

• Has a host of third-party libraries that extends its use with high-level user interface

• Handles data quickly

• Dashboards with Bokeh server are dynamic and very fast

• Python centric

• Limited interactivity

• No 3D plotting

• Plotting code is more intense

• Styling graphs is more difficult

• Bokeh server: Server introduction — Bokeh 3.0.3 Documentation

• Can add custom Javascript for advanced/specialized cases

D3.js (D3.js - Data-Driven Documents (d3js.org))

Javascript

• Open source

• Free

• No, but must provide provide properly referenced path if you want D3 to work offline

• Requires a web browser

• Can form real-time graphs, charts, and extensive ways to visualize the data

• Open source and can be used with other Javascript frameworks (ie. Angular.js, Ember.js, or React)

• Can add own features to source code to accomplish your goals

• Handles DOM, HTML, CSS< SVG, and canvas → does not need other plug-in other than browser, does not need additional debugging

• Can create dynamic, real-time transformation by manipulating DOM elements into data visualization

• Works on data and is specialized and appropriate with data visualization functions in Javascript library

• Does not support older browsers → need to put static placeholder if users are on older browsers

• Cannot easily conceal original data → Challenging to use D3 if you want to hide data

• Does not generate predetermined visualizations for you.

  • Time-inefficient to generate D3 visualizations where an alternative source can do it faster and better

• Requires a web browser

• No cloud?

• Don’t think you can collaborate with others on D3??

• Data driven, gets data from arrays, CSV, XML, TSV, JSON, etc., and also API

• Should know/have:

  • Text editor

  • Web browser

  • HTML, CSS, DOM, Javascript

  • Web server

Pts.js

Javascript