Abstract: With the transformation of China's energy structure, the proportion of natural gas in primary energy consumption continues to rise. China's natural gas liquefaction and receiving station projects continue to launch, and the proportion of LNG in the natural gas consumer market keeps rising. In the traditional project analysis, the use of fixed natural gas prices for measurement does not reflect the changes in the consumer market. This paper provides a dynamic prediction model based on machine learning method. The regression prediction is carried out by analyzing the key parameters of the domestic market. In the analysis of natural gas imports in the past decade, the simulation prediction results are good, which can effectively predict the LNG imports in China’s coastal areas

Key words: machine learning, support vector machine/SVM, data analysis, liquefied natural gas/LNG