Classification of Spiral and Non-Spiral Galaxies using Decision Tree Analysis and Random Forest Model: A Study on the Zoo Galaxy Dataset

Abstract

Purpose: The goal of this research is to create a precise prediction model that can differentiate between spiral and non-spiral galaxies using the Zoo galaxy dataset. Decision tree analysis and random forest models will be used to construct the model, and various conditions within the dataset will be employed to classify the data accurately. The model's performance will be evaluated using a confusion matrix, and the probability of predicting spiral galaxies will be analyzed. The research will also investigate the differences in Total Power among signal types and identify Peak Frequency and Bandwidth values consistent across all signal types. This study is expected to provide important insights into galaxy classification and signal characteristics, specifically in the fields of astronomy and astrophysics.Methods: This study utilized the decision tree analysis research method to create a predictive model for identifying spiral galaxies using the Zoo galaxy dataset. The research approach focused on analyzing data before constructing a prediction model. The study did not involve random sampling, making it an observational study. Decision tree analysis was employed to classify galaxies into homogeneous groups, and a random forest model was used to classify galaxy types. This research provides insights into how decision tree analysis can be utilized to comprehend galaxy classification and can serve as a foundation for future research. To strengthen the conclusions, combining this research with other approaches such as experiments or random sampling can be considered.Result: This study developed a predictive model for classifying galaxies based on their Spiral type using decision tree analysis on the Zoo galaxy dataset. The model divided the data into specific groups based on certain conditions, and the results demonstrated exceptional accuracy of the random forest model in categorizing galaxy types. In addition, the study investigated various signal types in galaxies and found variations in Total Power, but consistent values for Peak Frequency and Bandwidth at 2 in all signals. These findings provide valuable insights into galaxy classification and signal characteristics, which could have practical applications in communication, signal processing, and analysis. The utilization of decision tree analysis and random forest models for galaxy classification and signal analysis represents an innovative approach in this field.Novelty: The novelty of this research lies in the new approach to categorizing galaxy types using decision tree and random forest models. Previously, the approach used to categorize galaxy types was through visual methods and observations via telescopes. This new approach provides a new and potentially more efficient way of processing galaxy image data, resulting in faster and more accurate categorization. Moreover, this research contributes to the development of signal analysis applications such as Total Power, Peak Frequency, and Bandwidth, which were previously only used in the fields of astronomy and astrophysics. However, they have the potential for wider applications in the fields of communication, signal processing, and analysis beyond astronomy