Rich data and inadequate knowledge can be effectively addressed with data mining technology. An essential component of data mining is cluster analysis, which includes concepts based on partition, hierarchy, density, grid, and model analysis. One common model-based clustering technique is the Levenberg-Marquardt method. It is an organic fusion of data mining and brain cognitive science. It is theoretically well grounded in our understanding of how the brain processes information. In order to optimize the network clustering data mining algorithm, this study employs the Levenberg Marquardt algorithm. Additionally, experiments are designed to validate the suggested Levenberg Marquardt method data mining clustering optimization algorithm. According to the experimental study findings presented in this paper, the Levenberg Marquardt data mining clustering optimization methods have mean values of 78.8 and 89.92 on four datasets, respectively, which are higher than the values of the other algorithms. The technique is also used in this research to examine the distribution of leftover oil. The cluster analysis of the flooding degree shows clear results that the algorithm has produced.