Using data mining technology (fuzzy mining technology), a reasonable and effective method to predict the lifespan of shovel cables is proposed. Shovel cables are expected to last approximately 2000 hours of operation. However, current lifespan range from 400 to over 1800 hours over an entire shovel fleet. The proposed approach can discover the correlation (i.e., the degree of fuzzy association) between a cable?s lifespan and operating variables. The degree of fuzzy association is defined based on the distribution of the variables for the lifespan and the concept of semantic proximity (SP) between two lifespan. In addition we adopt the inverse document frequency (IDF) weight function to measure the weights of the variables in order to superpose the association degrees. Given the proximity degree (PD) between two time-series, the time-series can be successfully classified using the fuzzy equivalence partition method. To implement the method, we introduce "growing window", "scaling", and approximate computation pruning techniques in order to reduce both I/O and CPU costs. Extensive experiments on real datasets are conducted, and the experimental results are analyzed thoroughly.