Method and system for predicting carpool matching probability in ridesharing

A system comprising one or more processors and one or more non-transitory computer-readable memories coupled to the one or more processors, the one or more non-transitory computer-readable memories storing instructions that, when executed by the one or more processors, cause the system to perform operations comprising:

• training a first machine-learning model based at least on training data, wherein the training data comprises, for each of a plurality of historical carpool requests, a pair of historical price adjustment multipliers applied to the historical carpool request, a rider-expected carpool matching probability of the historical carpool request, and a label indicating whether the historical carpool request was converted to a historical trip;
• generating, through the first machine-learning model, first output comprising a probability for a given carpool request to convert to a trip;
• training a second machine-learning model based at least on the first output of the first machine-learning model and a plurality of features of the historical carpool requests;
• generating, through the second machine-learning model, second output comprising a carpool matching probability for the given carpool request;
• constructing one or more Key Performance Indicator (KPI) models based on the second output and a pair of price adjustment multipliers;
• and
• determining optimal values of the pair of price adjustment multipliers based on an optimization model maximizing an aggregated value of the one or more KPI models.

A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

• training a first machine-learning model based at least on training data, wherein the training data comprises, for each of a plurality of historical carpool requests, a pair of historical price adjustment multipliers applied to the historical carpool request, a rider-expected carpool matching probability of the historical carpool request, and a label indicating whether the historical carpool request was converted to a historical trip;
• generating, through the first machine-learning model, first output comprising a probability for a given carpool request to convert to a trip;
• training a second machine-learning model based at least on the first output of the first machine-learning model and a plurality of features of the historical carpool requests;
• generating, through the second machine-learning model, second output comprising a carpool matching probability for the given carpool request;
• constructing one or more Key Performance Indicator (KPI) models based on the second output and a pair of price adjustment multipliers;
• and
• determining optimal values of the pair of price adjustment multipliers based on an optimization model maximizing an aggregated value of the one or more KPI models.