Consensus Model

KNIME Analytics Platform
1

 

 

Hi

I am working on Classification problems in Weka implemented in KNIME. I have developed some models from SVM, RandomForest and Decision Tree. These methods predicts quite similar to my dataset (prediction is nearly 72, 79, 80 % respectively), but i would like to combine these models into one model. I know, how to make these consensus model in EXCEL(manualy) but not in knime.

simple e.g. is 

Observation 1   Actual          SVM        RF         DT          Consensus

1                     Positive (P)       P         P           P               TP (All methods correctly classified)

2                     Positive (P)       N         P          P               TP (Majority of Method in correctly classified)

3                     Negative (N)     P         P          N               TN (Majority of Method in correctly classified)

4                     Negative (N)     N        P          P                TN (Majority of Method correctly classified)

5                     Negative (N)      P       P           P                FP (Majority of Method correctly classified)

Prediction %                            40      40        60              80

i wanted to make automatic consensus model out of developed models. I am quite new to knime and i dont have programing skill. Could anyone explain this how to make ? is there any node for this ?

Best regards

Nathan

 

2

For me it looks like if you can simply run each Predictor node independently on your data, apply the Scorer node, and concatenate the the results?