Good morning. My project involves the comparison of three different machine learning algorithms (RF, GBT, and KNN) in order to identify the one that best discriminates between active and inactive molecules for a specific enzyme. In my study, I also investigated how model performance changes when using different molecular fingerprints; therefore, each model was trained using five different molecular fingerprints. As a result, I obtained five ROC curves for each model. Is there a way in KNIME to overlay the five ROC curves (derived from different fingerprints) into a single plot? This would allow me to have only three plots, one per model, instead of fifteen.

@Giada_Turiano welcome to the knime forum. Yes this should be possible. [image] How to plot ROC curve for SVM predictor and compare will other models in one curve KNIME Analytics Platform I have attached a sample workflow for 3 scores and a ROC curve plot in two different ways. Maybe you could adapt that example. [roc1] [23] kn_example_roc3.knar (405.2 KB) [image] ROC Curve Examples – knime This workflow depicts utilizing the ROC Curve node to showcase the results of one classifier and two classifiers on a plot. We utilized the "Heart Attack Analy…

Overlay ROC curves for same model using different fingerprints.

KNIME Analytics Platform

mlauber71 December 21, 2025, 8:08am 2

@Giada_Turiano welcome to the knime forum. Yes this should be possible.

4 Likes