Parameters If you use the default parameters we provide, you don't need to set them. Just submit the file and wait for the result

The default (1) is equal to The first feature number at the beginning of the data set
The default( -1) is equal to The last feature number at the beginning of the data set
IFS strde(default 1)
save length after feature rank(default auto calculate by IFS)
select top n features to chart
feature rank method

classifier

If the task is completed, we will notify you by email
evaluate metric


( the example file click here)

Tips:

If your data is too large, We want you to leave an email. Other wise we recommend downloading the command-line version of MRMD3.0 from github and using it on a better configured machine.

Ref:If 5000<=samples/features<=10000 it may take hours or even days If the sample is greater than 10000

Parameters (see detail manual)
  • start_index:

      The first feature of the feature selection interval(defalut 1)

  • end_index:

      The last feature of the feature selection interval(defalut -1)

  • length:

      stride of feature selection(defalut 1)

  • top_n:

      select top n features to chart (defalut 15,violin plot, heatmap, stem map).

  • dimensions:

      set the number of features for feature selection, which is automatically calculated by default and does not need to be set

  • classifier :

      Classifier to use when inferring the number of features via IFS（default randomforest）

  • rank method :

      PageRank = 1,HITS:Authority = 2,HITS:Hub = 3,LeaderRank = 4,TrustRank=5

  • evaluate metric :

      evaluate metric(f1_score(default), accuracy, precision, recall,auc)

  • File Type:

      Supported data file formats are arff,libsvm or csv

  • If you have any problems,please contact me.

    heshida@tju.edu.cn