Method for defining intersections using machine learning

A method for defining a classifier for determining a geographic location of a portion of an intersection, the method comprising:

• defining a plurality of zones, each thereof comprising a corresponding sample intersection and each comprising a plurality of contiguous subzones;
• for each of the plurality of contiguous subzones of each zone of the plurality of zones, generating a plurality of features based on a subset of vehicle data indicative of vehicle operating conditions for a plurality of corresponding vehicles, the vehicle data including a subset of raw vehicle data or a subset of raw vehicle data and data interpolated therefrom;
• for each of the plurality of contiguous subzones of each of the plurality of zones, generating training data comprising the plurality of features to form training data;
• and
• using at least one machine learning technique with the training data to define a classifier for classifying a subzone as one of a portion of an intersection and not a portion of an intersection by:
• inputting training data into a machine learning algorithm,
• generating the classifier using the machine learning algorithm,
• evaluating a performance of the classifier,
• modifying at least one of the training data and a parameter of the machine learning algorithm, and
• continuing to:
• input training data into the machine learning algorithm,
• generate the classifier using the machine learning algorithm,
• evaluate a performance of the classifier, and
• modify at least one of the training data and the parameter of the machine learning algorithm, until at least one performance criterion is met.