BI & EDA Solutions

FOSS with SAAS

• metabase
  • toy self serve
    • deploy using singledocker container,jar
• apache-superset
  • toy self serve
    • deploy usingdocker-compose,k8s
• grafana
  • toy self serve
    • deploy usingdocker-compose,k8s
  • pro
    • dashboard can be exported and versioned as json
    • primary use case is system time series telemetry
  • con
    • cross version json compatibility is a joke
• redash
• kanaries rath
• KanariesPyGWalker
  • allows tableau eda experience inside Jupyter notebook
• bare-metal self serve
  • plotly +Dash +dbt for semantic layer
    • still uses pandas as final layer for plotting https://plotly.com/blog/polars-to-build-fast-dash-apps-for-large-datasets/
    • but nothing stopping using polars to do heavy data wrangling prior to convert to pandas for final plot.
  • bokeh +streamlit +dbt
  • bokeh seems somewhat limited to python whereas plotly is more widely adopted, go

Proprietary

• Qlik
• tableau

Proprietary that gets stuffed down everyone’s throat

• powerbi

Comparison

🖊 TODO

• refactor below into table

Interesting stuff thatpowerbitableau cant do, but come with Dash

• user edited table ( this allows form entry and calculator style interface).
• some custom visual exists inpowerbi appsource or marketplace but are all proprietory and paid.
  • Most likely require power platform (power app and dataflow as well)
• can be achieved with power app, but power app workflow free tier has 750 limit.
  • described as write back by powerbi and tableau