In this study, an actual hybrid GSHP system equipping with acooling tower, which is installed in an office building of Shaox-ing, is real-time monitored from Nov. 2012 to Mar. 2015. Basingon the monitoring data which have been processed by DM tech-nology, a post-evaluation procedure is conducted to evaluate theperformance of this GSHP system. The main operation problemsin this GSHP system are pointed out. Then this hybrid GSHP sys-tem is modeled by a DM algorithm, which is the most appropriatealgorithm selected out from six classical sorting algorithms underdifferent operation modes and each operation mode is respectivelymodeled by optimal algorithm. Follow that, the average long-termperformance of this hybrid GSHP system is predicted and evalu-ated under different operation modes by those data-driven models.Besides that, the relationship between long-term performance andshort-term performance of GSHP system is discussed and the majorfactors which directly affect the results of short-term assessmentare analyzed. Finally, the optimal time and number of days whenconducting a short-term monitoring with the purpose of predictingthe long-term performance of a GSHP system are provided.