Abstract: By utilizing the smart card data from Chongqing light rail system, the travel characteristics of light rail passengers are analyzed and a trajectory prediction algorithm based on Markov chain is proposed. In the algorithm, the next travel trajectory of a passenger is classified by Bayesian classification and then predicted according to the relationship between the passenger's last travel trajectory and her/his residence. Experimental results based on real datasets show that the algorithm outperforms LTMT, RNN and 2-MC on predicting passenger's next travel trajectory. Meanwhile, the algorithm is coded on Spark, a big data processing framework, which reduces its runtime.