Detecting overbooked flights and warning passengers, and applications thereof

Abstract

Described herein is a method, system, and non-transitory computer readable medium for monitoring transaction data for a user in a proactive and reactive mode for a detecting of a booking of a flight by the user in the transaction data, parsing flight detail parameters of the booking, and inputting parameters into a prediction model, such that if the likelihood of the flight being oversold exceeds a certain threshold, the user may be notified that the flight may be oversold.

What is claimed is: 
     
         1 . A computer-implemented method by one or more central computing devices for determining flight characteristics, the method comprising:
 retrieving values of parameters of a transaction record that corresponds to a ticket for a flight that a user has booked;   assigning weights to the parameters, the weights representing at least a portion of a machine learning model trained using transaction data indicative of overbooked or non-overbooked flights for a plurality of users;   feeding the values of the parameters as input values, into an input layer of the machine learning model, wherein each of the parameters are input features;   interpreting values of an output layer of the machine learning model to predict a likelihood of the flight becoming overbooked; and   based on the predicted likelihood of the flight being overbooked, notifying the user of the likelihood of the flight being overbooked.   
     
     
         2 . The method of  claim 1 , wherein the retrieving the values of the parameters of the transaction record is performed in response to determining that the user has booked the ticket for the flight based on analysis of a plurality of transaction records at a predetermined time interval. 
     
     
         3 . The method of  claim 1 , wherein the retrieving the values of the parameters of the transaction record is performed in response to determining that the user has booked the ticket for the flight based on analysis of a preconfigured number of new transaction records added into a transaction data repository. 
     
     
         4 . The method of  claim 1 , wherein the ticket for the flight is booked by the user is determined based on at least one of the parameters comprising:
 (a) a list of the user's past credit card transactions in a transaction data repository, (b) a list of commonly used flight aggregator sites and airlines in a flight data repository, and (c) a list of common user form-inputs associated with airline purchases in the transaction data repository.   
     
     
         5 . The method of  claim 1 ,
 wherein the machine learning model has additional inputs for predicting the likelihood of the flight being overbooked including one or more of:
 time of booking; 
 airline being booked; 
 name of a customer for whom the flight is being booked for; 
 number of passengers on the booking; 
 name of a customer who is purchasing the ticket for the flight; 
 frequent flyer mileage number; 
 weather history pertaining to the date of the booking; 
 historical weather data in dates proximate to the date of the booking within a predetermined time window; 
 number of customers in a transaction data repository that may have purchased or booked tickets for the same flight; 
 profile type of the user as a frequent or infrequent purchaser of tickets; and 
 profile type of the user as an economy spender or non-economy spender on tickets. 
   
     
     
         6 . The method of  claim 1 , wherein the notifying the user of the likelihood of the flight being overbooked is completed within a preconfigured time interval not exceeding twenty-four hours from a time of booking the ticket for the flight by the user. 
     
     
         7 . The method of  claim 1 , wherein the notifying the user of the likelihood of the flight being overbooked further comprises notifying the user via an email, a short messaging service (SMS), or a web browser extension. 
     
     
         8 . The method of  claim 1 , wherein the parameters comprise a parameter for weather of a connecting city of the flight for a date the flight is to occur, a parameter for available seat capacity, and a parameter for a number of days until departure. 
     
     
         9 . The method of  claim 1 , wherein the machine learning model is a multi-layer neural network machine learning model. 
     
     
         10 . The method of  claim 1 , wherein a greater weight is assigned to an input feature of weather of a connecting city of the flight than an input feature of a date of the flight, at the input layer of the machine learning model, when a value of the date of the flight is not within a predetermined first number of days from a holiday. 
     
     
         11 . The method of  claim 1 , wherein a greater weight is assigned to an input feature of a date of the flight than an input feature of weather of a connecting city, at the input layer of the machine learning model, when a value of the date of the flight is within a first number of days from a holiday. 
     
     
         12 . The method of  claim 1 , wherein the machine learning model indicates risk of overbooking to the user in one of a plurality of risk classes as an output, sorted by degree of risk, and wherein when the risk of overbooking is in a first sorted degree of risk, the user is notified that the flight is likely to be oversold, and when the risk of overbooking is in the a second sorted degree of risk, the user is notified that the flight is unlikely to be oversold. 
     
     
         13 . A system comprising a memory, a computer microprocessor coupled to the memory, and a flight data repository for storing flight data, configured to perform operations comprising: retrieve values of parameters of a transaction record that corresponds to a ticket for a flight that a user has booked;
 assign weights to the parameters, the weights representing at least a portion of a machine learning model trained using transaction data indicative of overbooked or non-overbooked flights for a plurality of users;   feed the values of the parameters as input values, into an input layer of the machine learning model, wherein each of the parameters are input features;   interpret values of an output layer of the machine learning model to predict a likelihood of the flight becoming overbooked; and   based on the predicted likelihood of the flight being overbooked, notify the user of the likelihood of the flight being overbooked.   
     
     
         14 . The system of  claim 13 , wherein the values of the parameters of the transaction record are retrieved in response to determining that the user has booked the ticket for the flight based on analysis of a plurality of transaction records at a predetermined time interval. 
     
     
         15 . The system of  claim 13 , wherein the values of the parameters of the transaction record are retrieved in response to determining that the user has booked the ticket for the flight based on analysis of a preconfigured number of new transaction records added into a transaction data repository. 
     
     
         16 . The system of  claim 13 , wherein the ticket for the flight is booked by the user is determined based on at least one of the parameters comprising:
 (a) a list of the user's past credit card transactions in a transaction data repository, (b) a list of commonly used flight aggregator sites and airlines in a flight data repository, and (c) a list of common user form-inputs associated with airline purchases in the transaction data repository.   
     
     
         17 . The system of  claim 13 , wherein the machine learning model has additional inputs for predicting the likelihood of the flight being overbooked including one or more of:
 time of booking;   airline being booked;   name of a customer for whom the flight is being booked for;   number of passengers on the booking;   name of a customer who is purchasing the ticket for the flight;   frequent flyer mileage number;   weather history pertaining to the date of the booking;   historical weather data in dates proximate to the date of the booking within a predetermined time window;   number of customers in a transaction data repository that may have purchased or booked tickets for the same flight;   profile type of the user as a frequent or infrequent purchaser of tickets; and   profile type of the user as an economy spender or non-economy spender on tickets.   
     
     
         18 . The system of  claim 13 , wherein the user is notified of the likelihood of the flight being overbooked is completed within a preconfigured time interval not exceeding twenty-four hours from a time of booking the ticket for the flight by the user. 
     
     
         19 . The system of  claim 13 , wherein the user is notified of the likelihood of the flight being overbooked via an email, a short messaging service (SMS), or a web browser extension. 
     
     
         20 . A non-transitory computer-readable storage device having instructions stored thereon that, when executed by at least one central computing device, causes the at least one computing device to perform operations comprising:
 retrieving values of parameters of a transaction record that corresponds to a ticket for a flight that a user has booked;   assigning weights to the parameters, the weights representing at least a portion of a machine learning model trained using transaction data indicative of overbooked or non-overbooked flights for a plurality of users;   feeding the values of the parameters as input values, into an input layer of the machine learning model, wherein each of the parameters are input features;   interpreting values of an output layer of the machine learning model to predict a likelihood of the flight becoming overbooked; and   based on the predicted likelihood of the flight being overbooked, notifying the user of the likelihood of the flight being overbooked.