Activity Identification of Androgen Receptor Ligand Binding Domain Using Voting-Based Ensemble Learning

In this work, efforts are made to construct a QSAR-based model, which employs computational methods for the detection of Activities to cut down on animal testing, time, and money in the first stages of medication development. To anticipate the behaviour of those pharmacological molecules that bind to the Androgen Receptor Ligand-Binding-Domain, an effective ensemble learning-based model is created. A methodology based on ensembles is suggested for categorizing activity. The 9442 medicinal compounds in the AR-LBD data set, and having a total of 1444 features, 372 of which are active and 9070 inactive. As a result, our dataset has a huge number of features and is quite unbalanced. After performing feature selection, the Smote algorithm was used to address the class imbalance issue. Our ensemble-based prediction model performs outstandingly in comparison to the legacy model. Performance analysis is shown in terms of accuracy, specificity, precision, F-score, and sensitivity, and the same is shown by the bar chart. Finally. The K-fold is carried out to assess the model's consistency.