A Study on Hotel Operation Performance Prediction Based on Feature Extraction and Random Forest Modeling

The page information displayed by the hotel online reservation platform directly affects the choice of consumers. Therefore, this paper proposes a method for predicting hotel business performance based on various models such as feature extraction, random forest, and neural convolutional network by utilizing online platform information. Firstly, the deep learning method was used to extract three types of data features: numerical, text and image, so as to compare and quantify the impact of various types of information presented on the online platform on the final sales. Then, the early fusion strategy is used to fuse the multi-modal features extracted in the previous stage to generate the multi-modal fusion feature vector. Finally, four classical machine learning algorithms were used for model training, namely: the Convolutional Neural Network model (CNN), the Random Forest model (RF), the Support Vector Machine model (SVM), and Gradient Boosted Tree model (GBT). The experimental results show that multimodal data fusion can significantly improve the prediction performance of the model and provide auxiliary suggestions for hotels in the operation process.