visualization_2018_workflow_plan

BCCN WeatherApp2018 Visualization toolbox
Main Workflow and quideline


User Interface: Our decision was creating a **web page** with interactive plots and visualization tools (further specification is still needed). This choice was reasoned by:

  1.) creating an android app needs a longer timescale to make it flexible
  2.) creating a GUI with a python package may look sketchy and prototype-like which would go against the main point of data visualization

What we definitely would like to achieve: A map where the GPS coordinate data is clearly mapped with any data collected from #analysis_2018

Ideas:

  1.) GPS data, station locations, translating from ID to map data
  2.) Temperature data, or any other raw data visualization on gived date from historical data
  3.) Analyzed data visualization, heatmap of prediction accuracy would be nice
  4.) Forecast data visualization
  5.) clickable map, interactivity

Workflow:

  1. plan the toolset, packages, toolboxes, compatibility with web design
  2. MAP
  3. getting example GPS data from actual DWD file and visualize it
  4. using colorcode, legend, whatsoever (“decoration”)
  5. “interactivity”, make it clickable
  6. … (further details depend on analysis and other parts)
  7. web design part (last but not least)
  
Toolboxes found:

  1.) Folium + GeoPandas (CHOSEN)
    Great for interactive map data which we can then add to a webpage. (Folium can handle GeoPandas data format dynamically)
    http://folium.readthedocs.io/en/latest/
  2.) mpld3 (CHOSEN)
    For every other kind of data visualization and plotting, creates interactive embeddable plots
    http://mpld3.github.io/


What can be done with Folium:
- clickable markers (vincent/vega markers) : https://folium.readthedocs.io/en/latest/quickstart.html#vincent-vega-markers
- time laps (HeatMapWithTime)

Problems with Folium:
- No outside interaction (maybe try to use a GUI, Bokeh or ipywidgets)