Fingerprint Bayesian Predictor

Predictor node to the Fingerprint Bayesian Learner node, assigning score values to test data. The input data needs to contain fingerprint descriptors as used in the corresponding learner. It computes a score for each input record by summing up the log values that are associated with the fingerprint on-bits (sum-of-logs). This corresponds to equation (6) in

Prediction of Biological Targets for Compounds Using Multiple-Category Bayesian Models Trained on Chemogenomics Databases, Nidhi Meir Glick, John W. Davies, and Jeremy L. Jenkins, J. Chem. Inf. Model., 2006, 46 (3), pp 1124–1133

This score represents the confidence of a record to belong to the same category as the target category (the attribute value that was selected in the Learner node). Additionally, the node allows the user to append a crisp class prediction. This prediction is done by comparing the computed score to a threshold, whereby the threshold can be either be fixed or a value derived from the model. Details are described below.

Options

Append Crisp Class Prediction: If selected, the output will contain a column containing a class label. It's the target category if a record's score is larger or equal to a threshold (see below). It's some other label (non-target) if it' below the threshold.
Use threshold from model: This threshold is derived from the training data. It's the score value on the leave-one-out predicted training data that minimizes the sum of errors on the target class and the non-target classes. This value is part of the model (you can see the value by inspecting the model output).
Use fix threshold: A numeric value that is used as threshold.
Use as label for non-target predictions: If the model is specific for one particular target class (which currently is always the case as there is no multi-target fingerprint bayesian node) the class prediction for non-target classes needs to be assigned.
Default label from model: The model may contain a label of a non-target class. This is the first non-target class in the class column of the training data (good default if you have a 2-class problem).
Fix label: A custom label. Leave the field empty to use missing values.

Input Ports

: The fingerprint model (output of the "Fingerprint Bayesian Learner").
: The data to predict.

Output Ports

: The input data with class prediction appended.

Popular Predecessors

Popular Successors

Views

This node has no views

Workflows

Developers

You want to see the source code for this node? Click the following button and we’ll use our super-powers to find it for you.

Installation

To use this node in KNIME, install the extension KNIME Base Chemistry Types & Nodes from the below update site following our NodePit Product and Node Installation Guide:

v5.4

A zipped version of the software site can be downloaded here.

Plugin provider: KNIME AG, Zurich, Switzerland

Plugin version: 5.4.0.v202410111128

On NodePit since: 2024-12-06

Last update: 2024-12-21

KNIME versions: Since v3.6

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!