Highest Rated Posts

Telcontar120

February 5

Re: ID Role

All special attributes (including id) are not included as predictors in modeling. You can't have more than one attribute with the role of id in the dataset at one time---if you try to do this, it will "unset" the prior id.  So the only way to have multiple id columns is to give the additional id columns custom special role names (which you can always do) such as id2 and id3.  

IngoRM

February 4

Re: Automatic feature engineering: results interpretation

Hi, On questions 1 and 2: I think @lionelderkrikor is right, there will be tiny differences which is the reason why the original one was picked.  In fact, if the accuracy would be exactly the same, the Pareto optimal result would be the feature set with 62 (since it dominated the other one). I would recommend trying a balance factor of 0.7 or 0.8 and you will likely get one of the results with 59 or 62 features.  In my own experience, I often do not want to optimize accuracy really with feature selection but get similar accuracy with smaller data sets (which makes training and scoring faster and the model less complex and more robust).  So I often end up picking a model in the 0.7 to 0.95 range... Question 3: Yes, most of them will contribute.  Obviously that depends a bit on the number of individuals during the optimization.  You can make the settings manually (increase the number of individuals to 100 or so but reduce the number of generations, otherwise it may run for ages...).  This will likely give you a more complete picture if really (almost) all features are necessary. Question 4: The best idea is to ignore the graph altogether :-)  We actually considered to remove it.  It was kind of useful while we have been tweaking some of the optimization heuristics, but it really is not that interesting in my opinion for users.  Anyway, here is what the chart tells you: for each generation, it tells you the highest and the lowest error rate of all individuals of this generation's population.  Same is true for the complexities: it shows the highest and the lowest complexity for the generation's population.  Since there are more individuals in each population than the Pareto-optimal solutions shown on the right, there does not have to be a 1:1 relationship between both charts. Towards the end of the optimization, however, there typically is one.  Have a look at the 4 points at the far right end (last generation).  The best error is 0.408 (dark orange) and the worst one is 0.465 (light orange).  Those are the points at the left and the right end of the Pareto front on the right.  At the same time, you can see the highest complexity is 142.  This is shown in dark blue which at the end of the optimization typically corresponds to the best performing feature set (which is why they share the dark color).  And last but not least there is the lowest complexity (50, light blue).  Which at the end of the run typically corresponds with the worst feature set with respect to error rate, hence the light color for both. Hope this helps, Ingo

IngoRM

February 1

Re: negative values with density plot

Hi @msacs09, I would like to second that from a modeling perspective the function is what it is and even if no negative values are part of your data, still the used distribution will cover both sides of the average value equally.  If you want to avoid this, you may want to try Kernel Naive Bayes instead. However, if you simply would like to get a distribution curve like yours, the upcoming RapidMiner release will offer a new (and very cool!) new visualization framework where you can easily create such a chart yourself (see below).  You can try it yourself already with the public beta version available here: http://static.rapidminer.com/rnd/html/rapidminer-9.2-preview.html Hope this helps, Ingo

Marco_Boeck Team Lead Software Engineering

January 23

Re: What will happen with RM's JRE version?

Hi, We've ran our full tests against the latest OpenJDK 8 build ~2 months ago, and everything was fine. That's not to say there may be some hidden problem somewhere in some case deep down, but we did not find any. This is not an official endorsement, but I personally would not expect any troubles really. In the past there was a problem with H2O and OpenJDK, because the H2O library is the biggest collection of hacks I ever saw, but I think that is no longer the case. Regards, Marco

lionelderkrikor

January 9

Re: Discretization before or after Feature Selection?

Hi @green_tea, I would say that Discretize the data must be performed before Feature Selection on the training set :  Don't forget to apply the same pre-processing step(s) to your test set... Hope it helps, Regards, Lionel

yyhuang

January 7

Re: Random Forest - Attribute Importance

Hi @asav_yu, have you run the model simulator in Auto Model? If you try "simulator" operator on your random forest trees, it will show you how the input "customer income" would affect the prediction interactively.. https://rapidminer.com/products/auto-model/ &lt;?xml version="1.0" encoding="UTF-8"?&gt; <output> </output>

Latest Posts

RapidMiner Community

Discussion List

• Can't find a shortcut icon in linux mint

  Responded But No Solution 23 views 4 comments 0 points Most recent by DanPayne 6:46AMHelp
• confusion matrix in rapidminer for clustering

  Responded But No Solution 30 views 5 comments 0 points Most recent by SamiRami 4:22AMHelp
• How to remove near duplicates e.g egg and eggs

  Responded But No Solution 8 views 2 comments 0 points Most recent by rfuentealba 12:12AMHelp
• Is there a place to find guidance for a college class assignment?

  Responded But No Solution 11 views 2 comments 0 points Most recent by rfuentealba February 17Help
• Confusion Matrix for Analyze Sentiment Operator

  Responded But No Solution 43 views 4 comments 0 points Most recent by IngoRM February 17Help
• Call a process

  Responded But No Solution 22 views 1 comment 0 points Most recent by IngoRM February 17Help
• What are the languages supported by Apache Spark for developing big data applications?

  Responded But No Solution 12 views 1 comment 0 points Most recent by rfuentealba February 16Help
• Performance = "unknown

  Responded But No Solution 22 views 2 comments 0 points Most recent by Telcontar120 February 16Help
• Hello

  New Question 28 views 0 comments 0 points Started by nazic_2015 February 16 Help
• Model exporting into other platforms

  Solution Accepted 40 views 4 comments 0 points Most recent by kypexin February 16Help
• How to calculate Correlation value on Auto Model?

  Solution Accepted 46 views 7 comments 0 points Most recent by angelotheus February 15Help
• Why confidence value is infinity?

  Solution Accepted 45 views 3 comments 0 points Most recent by lionelderkrikor February 15Help
• How to get Meta Data with the Data to Documents and Process Documents operators

  Solution Accepted 27 views 6 comments 0 points Most recent by dramhampton February 15Help
• Importing textdata from csv files

  Responded But No Solution 42 views 4 comments 1 point Most recent by sgenzer February 15Help
• extract keywords from a collection of text files(.txt) and obtain the number of occurences.

  Responded But No Solution 26 views 3 comments 0 points Most recent by Telcontar120 February 15Help
• Hive connection error

  Responded But No Solution 48 views 4 comments 0 points Most recent by Atekka February 15Help
• Calling Server webservice from a form on a webpage

  Responded But No Solution 14 views 6 comments 0 points Most recent by sgenzer February 15Help
• PDF Table extraction into data

  Solution Accepted 40 views 4 comments 1 point Most recent by robin February 15Help
• Design Question on join and Issue with replace missing value series operator

  Solution Accepted 62 views 4 comments 0 points Most recent by tftemme February 15Help
• Save regex capturing group for use later

  Solution Accepted 33 views 2 comments 0 points Most recent by mdale9 February 15Help
• How to import MS Access databases to Rapidminer?

  Responded But No Solution 34 views 2 comments 0 points Most recent by rfuentealba February 14Help
• Set MongoDB Instance with a macro, is it possible?

  Solution Accepted 32 views 4 comments 1 point Most recent by christian_ploch February 14Help
• Cross Validation operator model output

  Solution Accepted 35 views 4 comments 0 points Most recent by varunm1 February 14Help
• RapidMiner Course Videos from Universidad San Pablo CEU

  13 views 0 comments 1 point Started by sgenzer February 14 Knowledge Base
• Error Code 26

  Responded But No Solution 51 views 6 comments 0 points Most recent by Marco_Boeck February 14Help
• com.rapidminer.operator.UserError: The dummy operator Execute Python (replacing python_scripting:exe

  Responded But No Solution 56 views 4 comments 0 points Most recent by hughesfleming68 February 14Help
• work with text

  Responded But No Solution 27 views 2 comments 0 points Most recent by kypexin February 14Help
• Homework

  Responded But No Solution 31 views 2 comments 0 points Most recent by SGolbert February 14Help
• How to create dummy variable intervals, such as for times?

  Responded But No Solution 22 views 2 comments 0 points Most recent by kypexin February 14Help
• While exporting to Excel i need to replace "?" symbols with blanks

  Responded But No Solution 20 views 2 comments 0 points Most recent by kypexin February 14Help