Conducted by: Fabrizio Rossi, Professor of Operations Research, Department of Information Engineering, Computer Science and Mathematics, University of L'Aquila, Italy
• Developed a comprehensive machine learning pipeline for outdoor image classification using a dataset of 2,310 instances with 20 features across 7 categories (brickface, sky, foliage, cement, window, path, grass)
• Implemented data preprocessing techniques, including duplicate removal and statistical analysis, reducing the dataset size by 19.3% for improved model efficiency
• Applied unsupervised learning (K-Means clustering) and supervised learning algorithms, including Support Vector Machines (SVM) and Gradient Boosting Classifier (GBC)
• Optimized SVM performance using GridSearch hyperparameter tuning, achieving 96% accuracy compared to a 94% baseline
• Conducted comparative analysis demonstrating SVM superiority over GBC (96% vs 87.5% accuracy) through confusion matrices and classification reports