Anticipating user dissatisfaction via machine learning
A server of a transportation network determines characteristics of trips provided to users, as well as the usage of the services by users. Using the determined characteristics, the server trains a model that, for a given set of trips for a user, estimates a degree of likely user dissatisfaction with the trips. Based on the estimated degree of user dissatisfaction, the system can estimate user dissatisfaction in real time, directly after completion of a trip, and can take remedial actions should the user be estimated to be likely dissatisfied.
1. A computer-implemented method comprising:
determining characteristics of trips of a plurality of users of a transportation service;
determining, after completion of the trips, a set of users of the plurality of users that have ceased using the transportation services for at least a threshold period of time after the completion;
training a prediction model based on trip sequences associated with users who have ceased using the transportation services for at least the threshold period of time, the training comprising deriving, for each trip of a plurality of trips of trip sequences of the users who have ceased using the transportation service, trip features comprising:
divergences between estimated time of arrival and actual time of arrival, and
at least one of: data indicating whether a driver associated with the trip canceled the trip, or data indicating whether a user associated with the trip canceled the trip;
obtaining a first user dissatisfaction score for a first user using the prediction model and a recent set of trips for the first user;
determining that the first user dissatisfaction score indicates that the first user is dissatisfied with the recent set of trips; and
responsive at least in part to determining that the first user dissatisfaction score indicates that the first user is dissatisfied with the recent set of trips, performing a remedial action with respect to the first user.
2. The computer-implemented method of claim 1 , wherein the trip features comprise prices charged for the trips.
3. The computer-implemented method of claim 1 , wherein the trip features comprise trip waiting times.
4. The computer-implemented method of claim 1 , wherein the trip features comprise distances from users to trip commencement locations.
5. The computer-implemented method of claim 1 , wherein the trip features comprise degrees of congestion of the trips.
6. The computer-implemented method of claim 1 , wherein the trip features comprise ratings of drivers of the trips.
7. The computer-implemented method of claim 1 , wherein the trip features comprise express feedback about the trips from the users.
8. The computer-implemented method of claim 1 , further comprising computing a value estimate for the first user, wherein performing the remedial action with respect to the first user is responsive at least in part to the value estimate for the first user.
9. The computer-implemented method of claim 8 , wherein determining the value estimate for the first user comprises computing a function of average number of purchases by the first user and an average purchase price of purchases by the first user.
10. The computer-implemented method of claim 1 , wherein the remedial action comprises sending a promotion or rebate to the first user.
11. A non-transitory computer-readable storage medium storing instructions executable by a processor, the instructions comprising:
instructions for determining characteristics of trips of a plurality of users of a transportation service;
instructions for determining, after completion of the trips, a set of users of the plurality of users that have ceased using the transportation services for at least a threshold period of time after the completion;
instructions for training a prediction model based on trip sequences associated with users who have ceased using the transportation services for at least the threshold period of time, the training comprising deriving, for each trip of a plurality of trips of trip sequences of the users who have ceased using the transportation service, trip features comprising:
distances from users to trip commencement locations, and
at least one of: data indicating whether a driver associated with the trip canceled the trip, or data indicating whether a user associated with the trip canceled the trip;
instructions for obtaining a first user dissatisfaction score for a first user using the prediction model and a recent set of trips for the first user;
instructions for determining that the first user dissatisfaction score indicates that the first user is dissatisfied with the recent set of trips; and
instructions for, responsive at least in part to determining that the first user dissatisfaction score indicates that the first user is dissatisfied with the recent set of trips, performing a remedial action with respect to the first user.
12. The non-transitory computer-readable storage medium of claim 11 , wherein the trip features for the users comprise prices charged for the trips.
13. The non-transitory computer-readable storage medium of claim 11 , wherein the trip features comprise trip waiting times.
14. The non-transitory computer-readable storage medium of claim 11 , wherein the trip features comprise divergences between estimated time of arrival and actual time of arrival.
15. The non-transitory computer-readable storage medium of claim 11 , further comprising computing a value estimate for the first user, wherein performing the remedial action with respect to the first user is responsive at least in part to the value estimate for the first user.
16. A computer-implemented method comprising:
determining characteristics of trips of a plurality of users of a transportation service;
determining, after completion of the trips, a set of users of the plurality of users that have ceased using the transportation services for at least a threshold period of time after the completion;
training a prediction model based on trip sequences associated with users who have ceased using the transportation services for at least the threshold period of time, the training comprising deriving, for each trip of a plurality of trips of trip sequences of the users who have ceased using the transportation service, trip features comprising:
distances from users to trip commencement locations, and
at least one of: data indicating whether a driver associated with the trip canceled the trip, or data indicating whether a user associated with the trip canceled the trip;
obtaining a first user dissatisfaction score for a first user using the prediction model and a recent set of trips for the first user;
determining that the first user dissatisfaction score indicates that the first user is dissatisfied with the recent set of trips; and
responsive at least in part to determining that the first user dissatisfaction score indicates that the first user is dissatisfied with the recent set of trips, performing a remedial action with respect to the first user.