Enhancing Flower Type Classification Using K-Nearest Neighbors (KNN)
Year: 2025
Type: article
Abstract: This research uses a custom-generated dataset to implement the K-Nearest Neighbors (KNN) algorithm for flower classification. The dataset that has been taken for this research consists of 1,000 samples with features including petal and sepal dimensions and tilt angle. Data preprocessing techniques have been employed to improve and clean the manually generated dataset and optimize the performance of the classification model. A 10-fold cross-validation approach has been applied to evaluate the accuracy of classification algorithms and analyze the impact of different hyperparameters on the performance of the KNN algorithm. This study highlights the importance of the feature selection method, data normalization, and k-value optimization in achieving higher accuracy in the classification process. Furthermore, this research focuses on the challenges encountered during model training, such as class imbalance and overfitting. This study also provides perceptions of enhancing KNN's efficiency in multi-class classification tasks and establishes its potential for real-world applications in automated flower identification and similar pattern recognition problems.
Authors: Prit Senjaliya, Khushi Prajapati, Avani Vagadiya
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4413067962