Method for predicting trip purposes based on input topics utilizing a purpose prediction model

A system, comprising:

• one or more processors;
• and
• a memory storing instructions that, when executed by the one or more processors, cause the system to:
• gather travel data by an application executing on a mobile device;
• send the travel data to a remote server;
• receive a recommended trip purpose from the remote server, wherein:
• the recommended trip purpose was determined at the remote server based on the travel data using a purpose prediction model trained to output trip purposes based on input topics;
• a topic was determined at the remote server based on the travel data using a topic prediction model that was trained, using a plurality of topics and one or more features derived from each of a plurality of trip records associated with the plurality of topics, to output topics based on input trip records;
• and
• the topic was provided as an input to the purpose prediction model;
• display the recommended trip purpose in a user interface of the application;
• and
• provide a first user interface element configured to cause a trip record to be labeled with the recommended trip purpose by a single user interface interaction with the first user interface element.

A method for recommending trip purposes in a user interface of an application, the method comprising:

• gathering travel data by an application executing on a mobile device;
• sending the travel data to a remote server;
• receiving a recommended trip purpose from the remote server, wherein:
• the recommended trip purpose was determined at the remote server based on the travel data using a purpose prediction model trained based on a topic model to output trip purposes based on input topics;
• a topic was determined at the remote server based on the travel data using a topic prediction model that was trained, using a plurality of topics and one or more features derived from each of a plurality of trip records associated with the plurality of topics, to output topics based on input trip records;
• the topic model represents words associated with a plurality of topics based on statistical analysis of a plurality of trip purposes and was built by using an unsupervised generative technique to determine the plurality of topics from the plurality of trip purposes based on probability;
• and
• the topic was provided as an input to the purpose prediction model;
• displaying the recommended trip purpose in a user interface of the application;
• and
• providing a first user interface element configured to cause a trip record to be labeled with the recommended trip purpose by a single user interface interaction with the first user interface element.