To address the issues of low positioning accuracy and the risk of overfitting when the random forest (RF) algorithm is used for indoor visible light positioning, a sparrow search algorithm(SSA) optimized RF algorithm for indoor visible light positioning based on sine population mapping(SPM) and Cauchy distribution(hereinafter referred to as SCSSA-RF algorithm) is proposed. Firstly, this algorithm establishes a fingerprint database using the received signal strength values and position coordinates. Then, it uses the global search capability of SCSSA to optimize key parameters of the RF algorithm, inputs the data into the optimal model for training, and averages the prediction results from the decision trees to obtain the predicted value for the target location point. The experimental results show that SCSSA-RF algorithm converges faster than the unimproved SSA-RF algorithm, the average positioning error of SCSSA-RF algorithm is 0.08 meters, with errors mainly concentrated between 0.05 to 0.1 meters. At a positioning error of 0.2 meters, the prediction accuracy of SCSSA-RF algorithm reaches 93%.