5 ms·

Show HN: Chartulo.us - Chart anything from the command line

45 points by dcraw 14y ago

3amOpsGuy 14y ago
Ahh cool, i see the attraction. I went through a phase of using the google chart API instead of GNUplot (you just wget a URL piping your data). Coupled with a big bash history it works fairly well. I've learned enough R to be dangerous. Ggplot2 looks gorgeous without trying. It's my current favourite torch for shining on perf and capacity issues (I dream of one day identifying a perf issues root cause with nothing more than the correlation function!)
anusinha 14y ago
You know, if you're already on the command line, gnuplot and github already provide ~similar functionality. I can do the same in ~the same number of keystrokes if things are aliased properly. Also, keeping things local is nice.
- dcraw 14y ago
  Hey, that's a good point. We're also interested in collaboration. How do you share your charts once you create them?
  - anusinha 14y ago
    Nothing fancy. Either a shared folder in my home-directory for users on the same machine, or email. Most of my work is separate individuals showing off their data rather than working together on one dataset. That could be a cool feature though--project groups or something.
- bradleyland 14y ago
  I use gnuplot to work with large datasets I extract from telco carrier logs (for client usage analysis). It's one of the few tools you can throw a 100,000 point dataset at and it graphs it in seconds. Try that with most scripting language graphing libraries, and you'll be there for a few minutes. The problem I have with gnuplot -- and maybe I'm the limitation, not the software -- is that I find it difficult to use, and the result of some input is inconsistent. Not to appear self-deprecating, but I really do think I might be the limitation here. Regardless of that, I often waste time trying to figure out why gnuplot is behaving as it is. I have no idea if chartulo.us is going to choke on my datasets, but I hope I'll be able to use it. I really could use a tool that is a little easier to use than gnuplot.
  - anonDataUser 14y ago
    Tableau does nearly instant plotting of millions of data points with a nice user interface; it may be worth checking out if your job depends on visualizing massive amounts of data.
olefoo 14y ago
It looks nice. If you're looking for an open alternative there is https://github.com/mikedewar/d3py https://github.com/mikedewar/d3py which plays very well numpy/pandas.
- dcraw 14y ago
  Ah, that's a cool project! We will check it out.
egonschiele 14y ago
Looks cool, I've been wanting something like this for a while. Also, your animated quickstart is annoying. Leave it as text please.
- seehafer 14y ago
  That's great. Can you give us some idea of what kinds of data you're looking to chart and what sort of ability you'd like out of a tool like this? Noted on the animation.
tkahn6 14y ago
This may be a dumb question but is this a startup or a project?
- hollerith 14y ago
  When (as in the current case) I don't see a way to download source code, I assume it is a startup or a project that will become a startup if it becomes sufficiently popular.
  - seehafer 14y ago
    That's right. Right now we're trying to get a handle on the interest in a tool like this.
q_revert 14y ago
another option/alternative is to use matplotlib, which is ideally integrated with ipython notebooks http://ipython.org/ http://ipython.org/, well worth a look if anyone's interested.. their integration with numpy/pandas etc is superb.. and on top of all the benefits of plotting with matplotlib, you get ipython and all its features for free... for example http://imgur.com/Le8px http://imgur.com/Le8px
Qworg 14y ago
At least from my perspective, a major problem is maintaining control of your information. How does Chartulo.us store the data? Where? Can it be removed immediately post retrieval?
dinkumthinkum 14y ago
The animation of the command line is a little annoying because it has all this cruft of the package installer messages. I got bored half way through and missed and had to replay. Just put the two command lines there and be done with it. :)
thronemonkey 14y ago
I use ggplot2 all day baby. http://had.co.nz/ggplot2/ http://had.co.nz/ggplot2/ It has a bit of learning curve but its an extremely flexible way to quickly visualize data in a lot of different ways to get a feel for what's going on in your dataset (you should probably already have some R familiarity, the language has its idiosyncrasies).
duaneb 14y ago
I would much prefer generating the graph locally and then uploading to the service - it's very rare that graphs I generate need to be shared rather than embedded in a paper.
- seehafer 14y ago
  OK. Are current tools (like gnuplot and the others mentioned in this thread) good enough for you or is there some room for improvement?
walrus 14y ago
Not that it really matters, but chartulous < data.csv http://partmaps.org/era/unix/award.html http://partmaps.org/era/unix/award.html
- dcraw 14y ago
  Nice, we already won an award! :)

45 points by dcraw 14y ago

Ahh cool, i see the attraction. I went through a phase of using the google chart API instead of GNUplot (you just wget a URL piping your data). Coupled with a big bash history it works fairly well. I've learned enough R to be dangerous. Ggplot2 looks gorgeous without trying. It's my current favourite torch for shining on perf and capacity issues (I dream of one day identifying a perf issues root cause with nothing more than the correlation function!)

anusinha 14y ago

You know, if you're already on the command line, gnuplot and github already provide ~similar functionality. I can do the same in ~the same number of keystrokes if things are aliased properly. Also, keeping things local is nice.

dcraw 14y ago
Hey, that's a good point. We're also interested in collaboration. How do you share your charts once you create them?
- anusinha 14y ago
  Nothing fancy. Either a shared folder in my home-directory for users on the same machine, or email. Most of my work is separate individuals showing off their data rather than working together on one dataset. That could be a cool feature though--project groups or something.
bradleyland 14y ago
I use gnuplot to work with large datasets I extract from telco carrier logs (for client usage analysis). It's one of the few tools you can throw a 100,000 point dataset at and it graphs it in seconds. Try that with most scripting language graphing libraries, and you'll be there for a few minutes. The problem I have with gnuplot -- and maybe I'm the limitation, not the software -- is that I find it difficult to use, and the result of some input is inconsistent. Not to appear self-deprecating, but I really do think I might be the limitation here. Regardless of that, I often waste time trying to figure out why gnuplot is behaving as it is. I have no idea if chartulo.us is going to choke on my datasets, but I hope I'll be able to use it. I really could use a tool that is a little easier to use than gnuplot.
- anonDataUser 14y ago
  Tableau does nearly instant plotting of millions of data points with a nice user interface; it may be worth checking out if your job depends on visualizing massive amounts of data.

olefoo 14y ago

It looks nice. If you're looking for an open alternative there is https://github.com/mikedewar/d3py https://github.com/mikedewar/d3py which plays very well numpy/pandas.

dcraw 14y ago
Ah, that's a cool project! We will check it out.

egonschiele 14y ago

Looks cool, I've been wanting something like this for a while. Also, your animated quickstart is annoying. Leave it as text please.

seehafer 14y ago
That's great. Can you give us some idea of what kinds of data you're looking to chart and what sort of ability you'd like out of a tool like this? Noted on the animation.

tkahn6 14y ago

This may be a dumb question but is this a startup or a project?

hollerith 14y ago
When (as in the current case) I don't see a way to download source code, I assume it is a startup or a project that will become a startup if it becomes sufficiently popular.
- seehafer 14y ago
  That's right. Right now we're trying to get a handle on the interest in a tool like this.

q_revert 14y ago

another option/alternative is to use matplotlib, which is ideally integrated with ipython notebooks http://ipython.org/ http://ipython.org/, well worth a look if anyone's interested.. their integration with numpy/pandas etc is superb.. and on top of all the benefits of plotting with matplotlib, you get ipython and all its features for free... for example http://imgur.com/Le8px http://imgur.com/Le8px

Qworg 14y ago

At least from my perspective, a major problem is maintaining control of your information. How does Chartulo.us store the data? Where? Can it be removed immediately post retrieval?

dinkumthinkum 14y ago

The animation of the command line is a little annoying because it has all this cruft of the package installer messages. I got bored half way through and missed and had to replay. Just put the two command lines there and be done with it. :)

thronemonkey 14y ago

I use ggplot2 all day baby. http://had.co.nz/ggplot2/ http://had.co.nz/ggplot2/ It has a bit of learning curve but its an extremely flexible way to quickly visualize data in a lot of different ways to get a feel for what's going on in your dataset (you should probably already have some R familiarity, the language has its idiosyncrasies).

duaneb 14y ago

I would much prefer generating the graph locally and then uploading to the service - it's very rare that graphs I generate need to be shared rather than embedded in a paper.

seehafer 14y ago
OK. Are current tools (like gnuplot and the others mentioned in this thread) good enough for you or is there some room for improvement?

walrus 14y ago

Not that it really matters, but chartulous < data.csv http://partmaps.org/era/unix/award.html http://partmaps.org/era/unix/award.html

dcraw 14y ago
Nice, we already won an award! :)