Find more posts tagged with
- Advanced Charts
- Export of Confusion Matrix
- Better decision tree exports
- Nicer inclusion of CSS in HTML Reports
- Addition of math formula layouts for models that can produce it
- Better Excel & PDF layout options
thanks for the update. Unfortunately, as far as I can see, it's not possible (or at least not trivial) to integrate a scripted ggplot/matplotlib solution with the Reporting extension, correct? If that is the case, I would have to look for an entirely different solution to generate the report surrounding the plots...
Thanks
Volker
Thanks for updating this thread. I too still await this sorely needed update.
I will say that I have spent the summer getting RapidMiner Server running on AWS and it has very nice chart features. And thanks to Tom Ott, there is a way to even use D3.js data visualizations within RM Server. If you send me your email, I can send you a link to a screenshot video I took to give you some idea of what this can look like. Also check out Tom Ott's YouTube video on D3.
Scott
It's currently possible to add both D3.js & advanced CSS & web styling into the HTML reports produced by the reporting extension so you can do quite a lot here already, it's just a bit manual.
I've also found that the need to get Advanced Charts into the reporting extension is a sorely needed feature.
So we're going to develop it.
I've got two guys who have joined our team who from mid-September will be working on updating the reporting extension as their first project with RapidMiner China (amongst their day-to-day). As soon as we have updates I'll let you guys know.
Features we're adding:
Oh, and one other thing might be useful: adding the possibility to expand macros in the HTML template...
Actually, if you have the image processing extension from Burgsys it's already available to you. You can connect the reporting extension to any image producing operator & choose Image as the reportable type. This works in all report outputs including PDF & Excel.
If you don't have the image extension then (you really should get it), but you can still do it with HTML reports and a combination of Macros, Write File & the Add Text operators.
Like many features of RapidMiner the Reporting extension is deceptively powerful and can do many things you might not expect it to.
I am going to ask the guys to make this a bit more usable and add some sample processes to demo some things.
You have given me a great idea, instead of just having the Reporting extension accept binary file inputs how about having it output binary files?
Currently the output options are 'write to disk' or 'write to repository', this serves well for static reports, but what if I want to operationalize the reports that I've built with RapidMiner Server? With a binary output I can design a process which produces a PDF report when a user clicks a download button on a RM Server web app.
Great News! You can definitly count on me in case you need a beta tester.
vwegert wrote:
This sounds rather exciting. I'm probably too recent a RapidMiner user to make any useful in-depth propositions, but what comes to mind - especially when taking into consideration Martin Schmitz' suggestion of using an external plotting library - would be an "Add Image to Report" operator. It could take some kind of binary stream with an associated mime type and then generate the file name, write the contents to the file and add the image tag to the HTML report (or whatever's necessary for other report formats). I haven't tried the Python/R intro capabilities yet, but if it's possible to render a plot there and feed it into the reporting output, it should improve the flexibility quite a bit - and it shouldn't be too hard to implement, since most of the functionality should already be present in other places.
Oh, and one other thing might be useful: adding the possibility to expand macros in the HTML template...
Something which comes into my mind: What's about interactive Charts? Like the stuff you can do with bokeh? (http://bokeh.pydata.org/en/latest/) ?
And additionally: It would be nice to append/integrate those charts into the Send Mail operator.
Best,
Martin
Although you can't have it display in PDF or Excel at this point, you can style your CSS so that it fits onto an A4 page when printing & then use a PDF printer after that.
By the way, there is an extension helping you to work with covariance matricies:
https://oldworldcomputing.com/
Sven, there's a demo license you can use it on small datasets; if you need bigger datasets then buy a license and support RM developers.
The pricing isn't going to break the bank so if it's a good product help them shift as many in volume as possible to keep them developing more.
Cheers
Sven
hi @uenge-san - so the update is that no work has been done on that extension for a while. Why? Few reasons...
- many users use BI platforms such as Tableau and Qlik, so we worked hard to support those extensions for smooth data xfer
- RapidMiner developer @land at Old World Computing built his own "Advanced Reporting" extension which we think is pretty impressive
That said, we do receive many requests for updating this extension. You can show your support here.
Scott
Hi @uenge-san,
how urgent would that development be for you? We are always offering custom development for RapidMiner and would be interested in getting this feature ourselves so that we could share the price for that. If someone else wants to have that feature we can open a hat and throw money in it
Greetings,
Sebastian
@land I think a new modern visualization workbench operator would be really helpful. I would buy that.
http://docs-beta.rapidminer.com/latest/studio/releases/changes-9.2.0.html
As it is based on HTML it would be awesome to include matplotlib or ggplot visualizations from R or Python scripting operators to the plots, as a lot of people know them from Jupyter notebooks.
PS: A create integration of the Jupyter was introduced in Knime just recently:
https://www.knime.com/blog/knime-and-jupyter
yes, we have plans to do so, but unfortunately no schedule yet. But admittedly a feature that I myself have been missing lately.
Best regards,
Marius