Evaluation of Pure-Electric Vehicle Retention Rate Based on TabTransformer

As China's NEV industry develops on a large scale with accelerated vehicle replacement, the speed of vehicle replacement has significantly accelerated, strongly promoting the development of the new energy second-hand vehicle market. This research targets low retention rates and hard valuation of pure electric vehicles. It proposes an evaluation method for pure electric vehicle retention rate using a deep learning model via variable screening, model construction (based on TabTransformer), and model performance evaluation. The research content and algorithm aim to be an important reference for entities like second-hand vehicle dealers, financial and insurance institutions, and consumers for evaluating pure electric vehicle retention rates.