Study for Iris Classification Based on Multiple Machine Learning Models

Abstract

Classification algorithms in machine learning aim to classify data into different kinds, which is important in data mining. It is important and challenging to select the classification model with high accuracy and efficiency. To address this, this paper compares and analyzes Multilayer Perceptron (MLP)with different machine learning methods, including K-NearestNeighbor (KNN), Support Vector Machine (SVM), Logistic regression, decision tree, and random Forest. All models are trained to classify different kinds of Iris to show further comprehension of different models. All the related models are evaluated with metrics called accuracy and confusion matrices. The experimental results show that the confusion matrices based on various models are similar but MLP overperforms other models in terms of accuracy. With sufficient nodes per hidden layer for backpropagation, MLP has a remarkable ability to analyze and provide projections. With this superiority, it is expected that the neural network can have better performance in classification soon. After understanding the strength and limitations of various classification algorithms, the optimization models are expected to be developed to better solve real-world problems.