Insurance Research Based on LSTM Model and BP Neural Networks
Peiqi Xu, Shihang Shen · 2024
This paper focuses on the application of machine learning algorithms to complex systems, with a particular emphasis on the use of LSTM (Long Short-Term Memory) models and BP (Back Propagation) neural networks. LSTMs, as a type of recurrent neural network, excel at working with time-series data, and are particularly well suited for predicting long-term trends and risks, whereas BP neural networks, with their error correction through multilayered feedback, perform well in classification and regression tasks. In this study, we combined algorithms such as entropy method, TOPSIS, LSTM model, BP neural network, DEA and gray prediction model to comprehensively analyze the effects of key variables in the system and accurately predict the efficiency in different scenarios. Subsequently, we constructed a multi-objective optimization model, aiming to balance the contradictions between different objectives in a complex system and enhance the effectiveness of decision-making. Further, we extended the application of machine learning to other areas, such as site selection analysis, and systematically evaluated potential development areas in six major cities in Guizhou Province using fuzzy comprehensive evaluation and AHP methods. The results show that Bijie, Guiyang and Anshun are identified as the best regions. Finally, this paper constructs a value assessment model based on AHP empowerment and fuzzy comprehensive evaluation, which solves the common problems of uncertainty and subjectivity in the assessment process, and effectively improves the accuracy and reliability of the overall assessment.