Protein subcellular localization is a well-known problem in the field of biological science. Finding out the location of protein inside the cell helps in understanding cell function, drug development, etc. This report presents a comparison between three standard machine learning classifiers: 1. Classification and Regression Tree (CART) 2. K-Nearest Neighbor (KNN) 3.Support Vector Machine (SVM) that are used to predict protein location. The Hold-out method is used for testing the classification models. The experiment is carried out on Yeast Dataset. The results are evaluated under four metrics: 1.Accuracy 2.Macro-average Precision 3.Macro-average Recall 4.Macro-F1 Score. According to results, SVM shows better Accuracy and Macro-average Precision value than two other classifiers, whereas CART shows better Macro-average Recall and Macro-F1 Score. Here R language by R-studio is being used to implement the coding phase of this whole project.
The goal of a new field of research called Explainable AI (XAI) is to make machine learning algorithms more transparent and comprehensible. This abstract looks at how XAI ideas are applied to the popular Random Forest algorithm, a powerful gradient boosting technique. While this specific model has proven to be a popular choice in many different industries due to its great prediction performance, understanding its inner workings might be difficult. This research examines the techniques and approaches used, such as force plots, dependence graphs, feature importance ranking, and SHAP (SHapley Additive exPlanations) values, to improve the interpretability of these particular decision tree models. We hope to demystify machine learning models with these XAI techniques, allowing users to get insights into how the algorithm makes predictions, discover potential biases, and make informed decisions based on model outputs. The combination of Random Forest and XAI shows potential for boosting model trustworthiness and facilitating model deployment in crucial applications where transparency and interpretability are critical. In this work the importance of explainability through the lens of emotional quotient features while preserving the sanity of the machine learning models is presented.