AN INTELLIGENT HYBRID MACHINE LEARNING FRAMEWORK FOR PERSONALIZED TRAVEL RECOMMENDATION AND EXPERIENCE RATING

Abstract

Travel planning has increasingly become complex due to the growing number of destinations, diverse user preferences, seasonal constraints, and the abundance of unstructured travel information. Conventional methods such as travel agencies, guidebooks, and informal recommendations often fail to deliver personalized and adaptive suggestions. To address these limitations, an intelligent hybrid machine learning (ML) framework was developed to provide personalized travel recommendations along with accurate experience rating predictions. The framework utilized a comprehensive dataset comprising user reviews, destination attributes, demographic details, and preference patterns. Data preprocessing was systematically performed, including label encoding (LE) of categorical features, term frequency–inverse document frequency (TF-IDF) vectorization of textual reviews, and standardization of numerical attributes to ensure model efficiency and consistency. Multiple machine learning algorithms were implemented, including Logistic Regression (LR), Support Vector Classifier (SVC), and a hybrid model combining Extra Trees (ET) and Multi-Layer Perceptron (MLP). The hybrid TREND-Net (Travel Recommendation using Extra Trees and Neural Network) model employs Extra Trees for feature selection (FS), identifying the most influential attributes, followed by a MultiLayer Perceptron classifier for predicting optimal travel destinations and corresponding experience ratings. This integrated approach enhanced both recommendation accuracy and prediction reliability. The performance of the models was evaluated using standard metrics such as accuracy (ACC), precision (PREC), recall (REC), and F1-score (F1), along with confusion matrix (CM) analysis, where the hybrid model demonstrated superior performance. An interactive graphical user interface (GUI) was developed to facilitate user interaction, enabling dataset management, exploratory data analysis (EDA), model training, and real-time personalized recommendations.