Ask HN: What do you use for basic data analysis, visualization, and graphing?

https://news.ycombinator.com/item?id=34875728

by transitivebs • 3 years ago

47 54 3 years ago

news.ycombinator.com

I often find myself with some JSON data that I want to visualize. I usually end up converting to CSV, uploading to a Google Sheet, and manually creating charts. BUT this is really time consuming and I find Google Sheets charts pretty difficult and painful for quickly exploring different views of the original data.

So what do you use for this type of thing?

I know python has lots of good utils for data wrangling & graphing, but I'd prefer a solution which is: no-code, gives me a bunch of common graph views I can quickly choose between, and that "just works" 99% of the time.

Thanks!

ghiculescu• 3 years ago

If you can connect it to the data source, Metabase is exactly what you want. https://www.metabase.com/

djbusby• 3 years ago

I love metabase. I'd also suggest to take a look at Apache Superset.

epgui2• 3 years ago

Personally I find Metabase lacks flexibility, but then again I think that’s just the predicament you’re stuck with in no-code/low-code.

ReDeiPirati2• 3 years ago

What do you mean with lacks of flexibility? I've used MB for 5y+ and I've never found myself limited by the tool as long as I was able to connect the data source, that's IMHO the main friction with this tool

transitivebs10• 3 years ago

Metabase looks amazing. Thanks!!

ReDeiPirati2• 3 years ago

This! As long as you can connect the data source you can go quite forward with this tool. It's probably the one that have the quickest learning curve.

n8henrie• 3 years ago

Consider adding "no code" to the title somewhere to save a click for the hundreds of people that are planning to cheerfully suggest jupyter + pandas / matplotlib / Altair / seaborn / R / etc etc.

transitivebs10• 3 years ago

Good call; I realized that after the fact and unfortunately can't edit the title anymore.

hprotagonist• 3 years ago

visidata, maybe?

__mharrison__2• 3 years ago

I use pandas. I'm pretty biased but I generally prefer to create things programmatically rather than drag and drop tooling. Especially if I need to do it in the future.

(I just made a course covering visualization w/ Pandas, Seaborn, Excel, Tableau, and a few others. My takeaway is that unless your data is good, you will need some preprocessing. Also, making good visualizations and tweaking them is difficult with code and no-code tooling. You need to figure out how to do the 20% of things (if you are even able to) in both code/no-code tools.)

gooseyman• 3 years ago

What is the name of your new course? I’m struggling to determine if it’s effective pandas or one of the others.

__mharrison__2• 3 years ago

This is a pending course that will come out through Coursera

flat-pluto• 3 years ago

I know you said wanted a no-code solution but in case you don't get a satisfactory answer try this out.

Earlier today there was a Show HN post[1] which showed how to visualize a Pandas dataframe (can come from CSV, JSON whatever). I tried it for basic tasks and it is pretty good. It's minimal code (<5 lines) - just reading the json and calling pygwalker in a Google Colab environment[2] or something. Something like this:

    import pandas as pd
    import pygwalker as pyg
    df = pd.read_json('{filename}.json')
    gwalker = pyg.walk(df)

Should be decent for most basic use-cases.

[1] https://news.ycombinator.com/item?id=34869244

[2] https://colab.research.google.com/

ellisv• 3 years ago

I use R and ggplot2 for most plotting/visualization but I’d recommend something like Apache Superset if you want no-code (although setup is still required).

wanderingmind• 3 years ago

You can use superset[0]. Its a Flask app that can connect to databases, read csv, json and create good plots

[0] https://superset.apache.org/

transitivebs10• 3 years ago

Looks very promising.

I think I'm looking for the AI-powered equivalent of this that's one level of abstraction higher. Apache projects are obviously super high quality, but I want to offload the cognitive load of thinking about the graph specifics to an ML algo that "just works" for the majority of use cases (and is tweakable after the fact).

aeontech2• 3 years ago

Datasette seems like it might be a good fit?

https://datasette.io/

kungfufrog• 3 years ago

+1 for Datasette, love it!

cldellow• 3 years ago

OP said they "find Google Sheets charts pretty difficult and painful for quickly exploring different views of the original data"

And that they want "a bunch of common graph views I can quickly choose between"

That doesn't sound like Datasette to me, although I'd be happy to be wrong -- how would you recommend someone achieve this in Datasette?

aeontech2• 3 years ago

Well, datasette has some simple exploratory viz built in, and is easy to extend if you want something more complicated. At least, I have been impressed with examples on the author's blog that come up here every so often.

I agree it's not the same as something like superset, but it is still a useful tool that some people might benefit from if they don't already know of it.

transitivebs10• 3 years ago

Checking it out now; thanks :)

acomjean• 3 years ago

Rstudio is a pretty nice user interface to the R language. datasets can be browsed like spreadsheet tables. ggplot2 is a great graphing tool. Used in science a lot by non programmers.

there is an online book too thats pretty decent: R for data science

https://r4ds.had.co.nz/

didgetmaster• 3 years ago

Here is a tool that will quickly load data from CSV, Json, or JsonLines formatted files using 'drag-and-drop' techniques. You can query it, analyze it, and display the information using a variety of charts.

https://www.youtube.com/watch?v=EHwBspzBDbs

Available for free download at https://www.Didgets.com

employee42• 3 years ago

Can't believe someone hasn't suggested Grafana[0] yet. It sounds perfect for your needs (although there is some coding required to make the queries).

[0] https://grafana.com/

neduma• 3 years ago

AWS Quicksight.

uptownfunk• 3 years ago

Why?

fbonawiede• 3 years ago

-1

rvrst• 3 years ago

Marple[0] is a pretty awesome tool for quick visualization and browsing through data.

[0] https://www.marpledata.com/

transitivebs10• 3 years ago

Looks very promising.

They seem to imply it's only for time-series data, but I like their marketing & UX so far. So many of the projects people link to are probably awesome, but if you don't nail the UX / DX, people bounce really quickly.

Thanks!

mutant• 3 years ago

https://www.spyder-ide.org/

Not no code, but can do simple things using something like nodes in blender.

kasperset• 3 years ago

Usually R tidyverse with GGplot2 does it for me. You could create shiny app and explore with "No code".

ies7• 3 years ago

Since you're already using Google sheet, fastest nocode dashboard available should be Google Data Studio.

transitivebs10• 3 years ago

Ahhh; didn't know about this. I guess it's called Looker Studio now? Will give it a try. Thanks!

naijaboiler• 3 years ago

It sucks

geophph• 3 years ago

I might suggest pandas + plotly express. Not no code, and dependent on your data structure, but if you can form it into a tidy data frame plotly express will let you easily customize into different chart types and styles from there

Gatsky• 3 years ago

Well, there is this chap on twitter who makes amazing charts to do with equities and macroeconomics (eg https://twitter.com/TimmerFidelity/status/161865104862910873...). Turns out he just uses Excel. Maybe that's not the answer you are looking for.

transitivebs10• 3 years ago

Still proves how far you can go when you master your tools. Great example && thanks.

ComputerGuru• 3 years ago

I used to do this with Matlab. Haven’t in a long time, but if you speak that lingua franca, it’s a great way to go. Every kind of chart, lots of flexibility, customizable, etc. but I think everyone has moved to Python or R now.

drewcoo• 3 years ago

The no code/low code poster child is Excel.

https://www.zdnet.com/article/office-excel-why-its-microsoft...

And there are many would-be Excels that do parts of what it does better.

Even if those contenders don't have a flight simulator yet.

https://excelunusual.com/a-first-flight-simulator-model-in-e...

shubhamjain2• 3 years ago

Shameless plug: This is exactly the problem the I am trying to solve with my app - TextQuery [1]. Creating even a basic graph means dealing with multiple tools. I wanted to create a simple app where you can import all common types of data, run SQL over it, and visualize it quickly.

[1] https://textquery.app/

transitivebs10• 3 years ago

Awesome; glad I helped validate the pain point && always happy to try and help fellow indie hackers.

btw adding a screenshot to the home page of the goal UX would help 1000x even if it's just a design mockup.

For my use case, I'd want to drag & drop a JSON file.

Thanks!

shubhamjain2• 3 years ago

Thank you! I would love to add screenshots as soon as I have something solid. Would it be okay to reach out to you for feedback on some of the things I have done?

transitivebs10• 3 years ago

I'm considering https://github.com/Kanaries/Rath, which seems to be an OSS version of Tableau. Has anyone used it for this type of thing?

noddingham• 3 years ago

Power BI Desktop is free and handles JSON files as a data source.

appleaday1• 3 years ago

Does Excel still freeze when trying to open huge .csv files?

b_mc2• 3 years ago

If I'm not trying to build a very specific graph or chart, and just exploring data I usually use either Rawgraphs or Sqliteviz. Rawgraphs is nice if you just want to swap visualizations out with smaller data as is, sqliteviz seems to handle much larger datasets and let's you use SQL if you want to change the resultset. Both seem to keep data local too and I know sqliteviz works offline, rawgraphs might too.

https://www.rawgraphs.io/

https://sqliteviz.com/

MilStdJunkie• 3 years ago

Not sure if this is "no code" exactly, but I use these from inside of Asciidoc every day, and I haven't seen them mentioned yet particularly. Asciidoc directives inside of a graph block are processed before the graph, so you can get conditional graphs in the output, or use the include directive to fetch outside graphics.

== Vega and Vega-Lite

Site:: https://vega.github.io/vega/

Sandbox:: https://vega.github.io/editor/#/examples/vega/airport-connec...

== PlantUML

Sandbox:: https://plantuml-editor.kkeisuke.dev/

Language Specs:: https://plantuml.com/sitemap-language-specification

georgewsinger• 3 years ago

Wolfram Mathematica.

victoriano• 3 years ago

Give a try to http://graphext.com . It’s no-code tool that at the same time is as powerful as a Jupyter Notebook

masteruvpuppetz• 3 years ago

Excel and that's all

msadowski• 3 years ago

In the robotics world many people use PlotJuggler (https://plotjuggler.io/) that, among other things, sports csv files.

iamcreasy• 3 years ago

You can write Trino rest client for a JSON endpoint that will allow you to query the endpoint using SQL and the result table can be push to any number of application for visualization.

Ask HN: What do you use for basic data analysis, visualization, and graphing?

Related Stories