Abstract: Wireline formation sampling technology represented by EFDT (enhanced formation dynamic tester), obtains formation fluid samples for fluid experimental analysis to derive parameters related to reserve calculation, which is widely used in oil and gas field exploration and development. In Bohai Oilfield, drilling operations generally use high-salinity saltwater mud, which leads to varying degrees of contamination of the original formation fluid due to factors such as mud filtrate invasion during sampling operations. On-site operations often rely on engineers' subjective experience to estimate the purity of pumped samples, lacking data support and decision-making basis. To address the issues of ineffective pumping waiting time and unqualified sample purity during sampling, big data statistical methods were applied to analyze fluid pumping time parameters from more than 20 sampling points in multiple exploration wells in Kenli Block. A fitting function was constructed to simulate the variation of fluid conductivity in water layer sampling with pumping time, thereby calculating the key parameter of formation water conductivity. Real-time fluid purity prediction models were separately established for water layers, oil-water layers, and oil layers. Practice shows that the error between this method and the actual sampling fluid purity is about 10%, greatly improving the efficiency of on-site sampling operations and having broad promotional and application value.