<![CDATA[Classifying agility levels presents challenges due to variations in team members’ personalities, roles, and undesirable behaviors. This study aims to enhance classification accuracy by comparing the performance of three algorithms: K-Nearest Neighbors (KNN), Random Forest, and Fuzzy-Particle Swarm Optimization (F-PSO) in classifying agility levels using simple feature scaling as part of the data preprocessing. Simple feature scaling is employed to ensure that all parameters are on the same scale, thereby improving the model’s effectiveness in learning classification patterns. F-PSO was selected for its ability to perform adaptive global search optimization within a fuzzy environment, while KNN and Random Forest serve as benchmarks. The study involved 160 participants from various Scrum teams to evaluate the effectiveness of these algorithms. The parameters considered included team members’ personalities (based on the Keirsey model), roles within the team, and the identification of negative behavior patterns (antipatterns). The results indicated that the F-PSO algorithm significantly outperformed KNN and Random Forest in terms of accuracy, improving from an average accuracy of 25% before optimization to 93.75% after applying F-PSO. This approach enables Scrum teams to identify and address obstacles affecting agility, facilitating earlier problem prediction and resolution, leading to more adaptive and effective teams.]]>
Copyrights © 2024
