A method, apparatus and computer program product are provided for predicting a split lane traffic pattern for a road segment. In this regard, first traffic data for an upstream road segment of the road segment is aggregated based on a distribution of speeds associated with location probe points representative of travel of vehicles along the road segment. Furthermore, second traffic data for a first downstream road segment of the road segment is aggregated based on the distribution of speeds associated with the location probe points for the vehicles. Third traffic data for a second downstream road segment of the road segment is also aggregated based on the distribution of speeds associated with the location probe points for the vehicles. Additionally, a machine learning model that predicts a traffic pattern is trained based on the first traffic data, the second traffic data and the third traffic data.