Rapidminer 5.0 Video Tutorial #11 – Pattern Recognition & Landmarking Plugin

I’m back to making new videos again, at least for a little while! This new video showcases the Pattern Recognition & Landmarking plugin that was unveiled at RCOMM 2010.

This plugin is fantastic!  It analyzes your data, ranks the best type of learners that should yield the highest accuracy, and then automatically constructs the process for you. Its so great that it helps answer one of the most often asked questions from my readers, “which learner should I use for my data?”

[flashvideo file=wp-content/uploads/2010/09/Rapidminer5-Vid11.mp4 /]

The video is NOT uploaded to my Youtube channel because its 13 minutes long.  Here’s the HQ video tutorial #11.

  • http://hype-free.blogspot.com/ Cd-MaN


    Just wanted to let you know that Youtube recently raised the time-limit from 10 minutes to 15 minutes, so you could upload the video to it.

    Best regards.

  • http://www.neuralmarkettrends.com Tom

    @CD-man: Thanks for the tip, I was there a few weeks ago and they still kicked me out for Vid #9 and Vid #10.

  • gfyang


    Thanks for the great video series. They are really helpful. Also, your English is quite understandable for non-native speakers.

    If possible, would you please give some more introductions about:
    (1), text mining & web mining;
    (2), how to develop our own operators/applications by JAVA using tools such as Eclipse.

    I once developed my applications using RM 4.5 by Eclipse, however, I find that RM 5.0 seems different and harder to use.

    Again, thank you very much for your great help.

    Sincerely yours,

  • http://www.neuralmarkettrends.com Tom

    @gyfang: Your welcome. I’m going to do a small video on text/web mining soon, but I can’t go into great detail in them because it would be cannibalizing the webinar revenue stream of the Rapid-I team.

    Regarding your question for operator/application development in Java, I don’t know how to do that but I should learn Java one day just for that reason. Perhaps you can ask your question at the Rapid-I forum?

  • gfyang

    Hi, Tom. I understand that. Thanks a lot.

  • wessel

    How does this plug-in relate to the No free lunch theorem?

    Of course the plug-in is seeing the actual data.
    But an extension of the PaReN plug-in could create a test and training set. If it would perform better then random on any problem, that would refute the no free lunch theorem.

    Wolpert will not be happy :P


  • http://pequaswans.blogspot.com/ ron mcewan

    Tom, I have given up trying to log into to the Neural Market Trends forum. So this seems to be the only way for me to forward any info. Sorry about that. Anyway I came up with what looks like a good way for doing “Pairs” analysis with RM. It was very simple. I read in an Excel file for the DJI and 30 stocks daily close and applied the “Mutual Information Matrix” The output generates a “Pairwise” plot which give an accurate picture of the “Pair” dependencies. Very Cool.

  • http://www.neuralmarkettrends.com Tom

    @Ron: That sounds pretty cool, perhaps you can share it on the MyCommunity site.

  • http://www.neuralmarkettrends.com Tom

    @Wessel: I’m trying to understand your comment so please forgive me if I didn’t understand you.

    The PaRen plugin works by reading in the data and then doing some quick optimizing on the specific learner and ranking it. So doesn’t the higher ranked learner have higher statistical significance?

    FWIW, I do get slightly different accuracy rates when I then take the PaRen process and create a test and training set using XVal.

  • http://pequaswans.blogspot.com/ Ron McEwan

    “perhaps you can share it on the MyCommunity site.” Good idea Tom, I will put it there now. I think this is an incredible tool for finding important relationships in data sets.

  • http://pequaswans.blogspot.com/ ron mcewan

    Now that RM integrates with R this may be useful

    Paper:Building Predictive Models in R Using the caret Package

    Building Predictive Models in R Using the caret Package

    The caret package, short for classification and regression training, contains numerous tools for developing predictive models using the rich set of models available in R. The package focuses on simplifying model training and tuning across a wide variety of modeling techniques. It also includes methods for pre-processing training data, calculating variable importance, and model visualizations. An example from computational chemistry is used to illustrate the functionality on a real data set and to benchmark the benefits of parallel processing with several types of models.


  • Thomas


    when trying this extension I got an error for also each sample data set like “Iris”.

    “Cannot read from XML stream, wrong format: com/rapidminer/operator/Annotations”

    I use RapidMiner 5.0 in its newest version on Mac OS. Could you help me?


  • http://blog.hiremebecauseimsmart.com stat arb

    You can go through this MUCH faster. It’s easy for me to pause a video or go back if I didn’t understand something but much harder to fast-forward on the Web.

  • http://www.neuralmarkettrends.com Tom

    @Thomas: Its really difficult to help you without seeing the process XML file or screenshots…

    @stat arb: of course I could but RM is a difficult subject to grasp for first time users so many appreciate my glacial pace.

  • John

    I noticed in your extension manager you also had the R extension. I’ve not been able to get the extension to work for me, I was wondering if you did and if so could you write a blow by blow explanation I could follow. I followed the rapid miner procedures(correctly i think) but its not working. I hope you can help.



  • John

    I just saw your recent post, http://www.neuralmarkettrends.com/2011/01/11/a-botched-r-plugin-installation-in-rapidminer-solution/, and am in the process of reading with interest.