Icon

SVM_​Workflow_​γ and σ fine-tuning

There has been no title set for this workflow's metadata.

This workflow shows how to utlize the parameter optimization methodology for varying the c-param and sigma-value for radio base function (rbf) in SVM ML application. KNIME node generally offers the sigma parameter for optimization while the c-param is available as a flow variable and is not directly visible. The approach has been adopted basis the approach mentioned for running SVM in Python as part of the Udemy course.

URL: 01-Support Vector Machines with Python.ipynb https://github.com/nilotpalc/Py_DS_ML_Bootcamp-Master/blob/434d3d3ac21c17a3de608186b854534183998df3/16-Support-Vector-Machines/01-Support%20Vector%20Machines%20with%20Python.ipynb

The parameter σ in the Radial Basis Function (RBF) kernel of Support VectorMachines (SVM) is related to the spread of the kernel and therefore the decisionregion.- Small σ - The support vector doesn't have a wide-spread influence. It onlyaffects the outcome for points very close to it. This leads to a more complexdecision boundary and possibly to overfitting.- Large σ - The support vector has a wide-spread influence, and it affects theoutcome even for points far away from it. This leads to smoother decisionboundaries and possibly to underfitting.It's important to note that in the context of SVMs in `sklearn` python codeapplication, the `gamma` parameter is typically used instead of `sigma`. `gamma` isdefined as 1/(2σ²), so a large `gamma` corresponds to a small σ, and vice versa.Therefore, when tuning hyperparameters, you're more likely to see `gamma` ratherthan σ.As with the `C` and `gamma` parameters, choosing the right value for σ (orequivalently, `gamma`) is crucial for the performance of the SVM model, and istypically done through hyperparameter tuning. Node 1Node 2Node 3Node 4Node 5Node 7Node 8Node 9Node 10Node 11Node 12Node 13Node 14Node 15Node 16Node 17Node 18Node 19Node 20Node 21Node 22Node 23Node 24Node 25Node 26Node 27Node 28 Python Script Table Creator Math Formula Column Expressions SVM Learner Table Row ToVariable Loop Start Math Formula Column Filter Column Filter Cross Joiner Partitioning SVM Predictor Scorer Column Expressions Row Filter Column Filter Loop End Pivot Python View Math Formula Math Formula Number To String Column Resorter SVM Predictor SVM Learner Scorer Variable Creator The parameter σ in the Radial Basis Function (RBF) kernel of Support VectorMachines (SVM) is related to the spread of the kernel and therefore the decisionregion.- Small σ - The support vector doesn't have a wide-spread influence. It onlyaffects the outcome for points very close to it. This leads to a more complexdecision boundary and possibly to overfitting.- Large σ - The support vector has a wide-spread influence, and it affects theoutcome even for points far away from it. This leads to smoother decisionboundaries and possibly to underfitting.It's important to note that in the context of SVMs in `sklearn` python codeapplication, the `gamma` parameter is typically used instead of `sigma`. `gamma` isdefined as 1/(2σ²), so a large `gamma` corresponds to a small σ, and vice versa.Therefore, when tuning hyperparameters, you're more likely to see `gamma` ratherthan σ.As with the `C` and `gamma` parameters, choosing the right value for σ (orequivalently, `gamma`) is crucial for the performance of the SVM model, and istypically done through hyperparameter tuning. Node 1Node 2Node 3Node 4Node 5Node 7Node 8Node 9Node 10Node 11Node 12Node 13Node 14Node 15Node 16Node 17Node 18Node 19Node 20Node 21Node 22Node 23Node 24Node 25Node 26Node 27Node 28Python Script Table Creator Math Formula Column Expressions SVM Learner Table Row ToVariable Loop Start Math Formula Column Filter Column Filter Cross Joiner Partitioning SVM Predictor Scorer Column Expressions Row Filter Column Filter Loop End Pivot Python View Math Formula Math Formula Number To String Column Resorter SVM Predictor SVM Learner Scorer Variable Creator

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: Nilotpal.Choudhury
Created at: 2023-07-24
On NodePit since: 2023-07-25
Last update: 2024-11-21
Created with KNIME version: v5.1.0
Tags: svmsupport vector machines

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

Try NodePit Runner!