Daten.Cafe

Prototype - Pre-Alpha!!!

Simple No-Code Browser-based Data Processing and Visualisation Tool

Daten.Cafe draws heavily from other low/no-code open-source tools, in particular Knime and Orange3. We adopted most of the element icon from Orange3. However, Daten.Cafe doesn't aim to be a full-blown data-processing plattform.

The goal of Daten.Cafe is to provide a (more or less) intuitive way to learn fundamental features of data, data processing and data visualization, thus enhancing data-literacy skills of everybody.

Protoype running at Daten.Cafe

Story at Medium

Instruction page should get you started. There is one example data-story available. Note: actions on nodes are started with a LONG click. Works on mobile too but less fun. You need to toggle between flow and charts with, yellow button.

Daten.Cafe aims to provide basic knowledge in data literacy in a simple way. Free access to the application and sharable results are essential concepts. The application is not mobile-first, but at least mobile-compatible, allowing learning and practicing of concepts on (modern) smartphones.

Visit the prototye to get more information on idea and current implementation status.

Framework and Libraries

Framework is Ionic version 6 with Vue version 3. Upgrade to Ionic v7 requires Typescript migration on several files. Initial tests promising but not complete yet,

Flow editor is cytoscape. Potential lightweight alternative vue-flow. Custom/simple solution should be possible, we don't need much magic here.

Data processing library is Danfo.js providing "Pandas"-like dataframes. Some Tensor functions available as well, not tested yet.

Orange3 widget symbols are used when possible.

XLSX

XLSX library is used to open Excel files and a older version (0.17.2) is included in Danfojs. There are security advisories for all XLSX versions up to 0.19.2 on Github and Sheetjs. However, the Danfojs version is not used to import files. Instead, we use a separate installation from Sheetjs, currently version 0.19.3. This security alert can therefore be ignored, unless you use the built-in XLSX version.

Data Processing

Node Types

To date (April 2023, tag 0.3x), the following elements are considered to be implmented. Only a (small but usefull) subset is available yet.

• Data Access

  • Load from Server
  • Load local
  • Start download via link (cord workaround)
  • Save local

• Single Item Operations

  • Statistics (Describe)
  • Select Columns
    • Rename option
  • Filter Rows
    • By Feature(s)
    • By Group
  • Reindex
  • To Timeseries
  • Resample
  • Simple Math Operations
    • Scale
    • Offset
    • Fill-NA (Impute)

• Dual Item Operations

  • Join on Index
  • Join on Feature
  • Math Operations
    • Add/Sub
    • Multiply
    • Divide (Ratio)
  • Correlation

• Geo Ops

  • Extract properties from features
  • Create properties from features => color for Chropleth maps

• Utilities

  • Load flow
  • Save flow
  • Export flow (png, pdf)

• Visualizations

  • Distribution /Histogramm
  • Table
  • Line chart
  • Bar chart
  • GeoMap
    • Points
    • Choropleth

Edge Types

Probaly only DataFrames and scalars.

Viz

Danfo has built-in plotting for dataframes using {Plotly](https://plotly.com/javascript/)

Maps will probably use Leaflet

Migration to Apache Echarts to be considered. Might be more lightweight and flexible

Icons

Basics: free fontawesome icon set

Individual SVGs and SVG-SPrites from here

If PNG icons needs, convert SVG like so (note: inkscape export is -e ?):

for i in *.svg; do inkscape -w 256 -h 256 $i -e basename $i .svg.png ; done for i in *.svg; do inkscape -w 256 -h 256 $i --export-type=png basename $i .svg.png ; done

Add border to icons like so

for i in *.png; do convert $i -bordercolor transparent -compose over -border 16 ../icons-bordered/$i; done

Note: remove the slash in front of $ ...

Development

Get started

1. Download repo
2. Install

npm i

3. Run dev server

npm run serve

4. Open Browser at http://localhost:8080/