Experimental and Comparative Analysis of Decision Tree, SVM, and MLP Algorithms for Raisin Dataset Prediction
Abstract
This study compares the performance of three machine learning algorithms, namely Decision Tree, Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP), for predicting outcomes on the Raisin dataset. The dataset comprises features related to raisin production, and the objective is to predict a binary outcome based on these features. The algorithms were evaluated using accuracy, precision, and recall metrics. The results demonstrate that the Decision Tree algorithm outperforms the others with consistent scores of 89.22% across all metrics. While SVM and MLP also yield strong performance, the Decision Tree algorithm emerges as the most reliable choice for predicting raisin production outcomes. These findings highlight the significance of algorithm selection and evaluation in machine learning tasks, providing insights for future research in agricultural forecasting and similar domains.
Keywords
Download Options
Introduction
Raisins are a concentrated wellspring of carbs and a nutritious bite, containing cell reinforcements, potassium, fiber and iron [4] Turkey is one of the nations that positions top on the planet's grape creation. Roughly 30% of the grapes delivered in Turkey are thought of as table, 37% as dried, 3% as wine and 30% as different items [5]. There are numerous uses of customary strategies for surveying and deciding the nature of food sources. In any case, these can be tedious and costly. Also, humancreated systems from customary techniques can be conflicting and more wasteful, as well as states of being, for example, weariness and, surprisingly, individuals' mental state of mind can influence the result of the work. These negative circumstances and issues are the primary purposes behind creating elective strategies to rapidly and unequivocally assess the fundamental highlights of items like raisins [8]. Machine vision framework is one of these elective strategies. Utilizing machine vision, it is feasible to extricate highlights from pictures and use them to gauge and assess the nature of different items [11]. Thus, while taking a gander at the examinations did as of late utilizing machine vision frameworks and picture handling methods on raisins from food items, it is seen that the items are analyzed as far as numerous actual highlights, for example, variety, surface, quality and size.
Conclusion
In conclusion, our experimental results highlight the comparative performance of Decision Tree, SVM, and MLP algorithms for predicting outcomes on the Raisin dataset. The Decision Tree algorithm exhibited superior accuracy, precision, and recall scores, indicating its effectiveness in classifying raisin production outcomes. While SVM and MLP algorithms also achieved commendable results, the Decision Tree algorithm stands out as the most reliable option for this specific dataset.
This study emphasizes the importance of algorithm selection and evaluation in machine learning tasks. However, further analysis and experimentation may be required to assess the algorithms' performance on other datasets or to explore other potential models that could provide even better predictive capabilities for the Raisin dataset. Overall, this research contributes to the understanding of the predictive modeling potential of various machine learning algorithms and serves as a foundation for further investigation in the field of raisin production forecasting or similar agricultural applications.