Optimizing Crop Yield Prediction: An In-Depth Analysis of Outlier Detection Algorithms on Davangere Region

TL;DR

This paper analyzes outlier detection algorithms to improve crop yield predictions in the Davangere region.

Contribution

The study evaluates six outlier detection algorithms for their effectiveness in optimizing crop yield predictions in a specific regional agricultural dataset.

Findings

Elliptic envelope showed improved performance in refining crop yield predictions.

Outlier detection significantly enhances the accuracy of crop yield prediction models in the Davangere region.

Abstract

Crop yield prediction is a critical aspect of agricultural planning and resource allocation, with outlier detection algorithms playing a vital role in refining the accuracy of predictive models. This research focuses on optimizing crop yield prediction in the Davangere region through a thorough analysis of outlier detection algorithms applied to the local agricultural dataset. Six prominent algorithms, including isolation forest, elliptic envelope, one-class SVM, iterative R, spatial singular value decomposition (SSVD), and spatial multiview outlier detection (SMVOD), are systematically evaluated. The study emphasizes the significance of accurate crop yield predictions in local agriculture and assesses each algorithm's performance using precision, recall, accuracy, and F1 score metrics. Elliptic envelope demonstrates its efficacy in handling the unique characteristics of the Davangere dataset. This method demonstrated improved performance in refining the crop yield prediction model by identifying and removing outliers, thereby contributing to more accurate predictions and optimized planning in the dynamic landscape of the Davangere region.

Figures