Ensemble Soft-Voting Model for Classification Optimization of Medicinal Plants Leaves

Growing public awareness of the value of going “back to nature’’ encourages the use of natural substances in pharmaceuticals. Chemical medications are thought to pose a great risk to the human body if used repeatedly, which has raised awareness of the importance of getting back to nature. Another issue is that not everyone in the community has access to information on therapeutic plants. Visible identification is thought to be ineffectual because of human mistakes, multiple sample testing, inadequate equipment, waiting times, and reading errors. Because of this, machine learning is crucial in finding patterns and plants more rapidly and accurately. The ensemble voting method is being used in this study to increase the accuracy of individual classifier findings. Additionally, this study analyzes various machine-learning techniques utilizing validation and test data. The process begins with feature extraction, which is followed by data normalization. Support Vector Machine (SVM), Random Forest (RF), and k-Nearest Neighbor (k-NN) are the three machine learning algorithms (individual classifiers) that are used to train the dataset. The prediction outcomes from the individual classifier algorithm are then entered into the ensemble voting method using soft voting. a person who uses a soft vote. 640 training data, 200 test data, and 160 validation data were divided randomly for the investigation. Individual classifier accuracy in this study was 91%, 94%, and 93% for SVM, RF, and k-NN, respectively, while accuracy for the ensemble soft-voting approach was 96%. Comparing the results of separate classifiers with the soft voting approach revealed an improvement in accuracy.