Hyperparameter Tuning and Pipeline Optimization via Grid Search Method and Tree-Based AutoML in Breast Cancer Prediction

Abstract

Automated machine learning (AutoML) has been recognized as a powerful tool to build a system that automates the design and optimizes the model selection machine learning (ML) pipelines. In this study, we present a tree-based pipeline optimization tool (TPOT) as a method for determining ML models with significant performance and less complex breast cancer diagnostic pipelines. Some features of pre-processors and ML models are defined as expression trees and optimal gene programming (GP) pipelines, a stochastic search system. Features of radiomics have been presented as a guide for the ML pipeline selection from the breast cancer data set based on TPOT. Breast cancer data were used in a comparative analysis of the TPOT-generated ML pipelines with the selected ML classifiers, optimized by a grid search approach. The principal component analysis (PCA) random forest (RF) classification was proven to be the most reliable pipeline with the lowest complexity. The TPOT model selection technique exceeded the performance of grid search (GS) optimization. The RF classifier showed an outstanding outcome amongst the models in combination with only two pre-processors, with a precision of 0.83. The grid search optimized for support vector machine (SVM) classifiers generated a difference of 12% in comparison, while the other two classifiers, naïve Bayes (NB) and artificial neural network—multilayer perceptron (ANN-MLP), generated a difference of almost 39%. The method’s performance was based on sensitivity, specificity, accuracy, precision, and receiver operating curve (ROC) analysis.

Automated Classification of Atherosclerotic Radiomics Features in Coronary Computed Tomography Angiography (CCTA)

“…Automated machine learning (AutoML) is a tool that automates the process of developing a machine learning model that has lately gained popularity. AutoML libraries come in a variety of environment including Auto-WEKA [ 29 ], Tree-based Pipeline Optimisation Tool (TPOT) [ 30 , 31 ], Auto-Sklearn, and others [ 32 , 33 , 34 ]. In this context, AutoML runs through a dataset and suggests the most optimum algorithm with the parameters set.…”

“…AutoML is the process of automating the application of machine learning to real-world situations from start to the end. AutoML can be explained mathematically as follows [ 31 ]: …”

Machine learning advances the integration of covariates in population pharmacokinetic models: Valproic acid as an example

“…We developed an easy-to-use model in the clinic by using only CYP2C19 genotypes and some noninvasive clinical parameters as predictors, and observed the influence of the omitted predictors on the performance of the proposed XGBoost model. Finally, we optimized the hyperparameters via the sklearn’s own grid search approach using the evaluation metric of MAE and tenfold cross-validation ( Radzi et al, 2021 ), and verified this simplified model after optimization in our independent external dataset, which consisted of 105 input-output data pairs retrospectively collected from our routine TDM practice according to guidelines of the Ethics Committee of the Affiliated Brain Hospital of Guangzhou Medical University approval ([2021] NO.027). The inputs to the external dataset were the same as those of the finally generated combined dataset with Single Dose, ALB, , and omitted.…”

An interpretable stacking ensemble learning framework based on multi-dimensional data for real-time prediction of drug concentration: The example of olanzapine

“…Third, the hyperparameters for each base model were adjusted to yield better predictive performance via the grid search method in scikit-learn. It evaluated the given hyperparameter combinations of the model by using “neg_mean_absolute_error” as the evaluation metric and 10-fold cross-validation ( Radzi et al, 2021 ). This technique allows inexperienced data scientists to acquire recommendations for tuning the parameters, but may be time consuming and inefficient in case of a large number of parameters.…”