Regression-tree compressed feature vector machine for time-expiring inventory utilization prediction
View Patent ↗This disclosure includes systems for regression-tree-modified feature vector machine learning models for utilization prediction in time-expiring inventory. An online computing system receives a feature vector for a listing and inputs the feature vector and modified feature vectors into a demand function to generate demand estimates. The system inputs the demand estimates into a likelihood model to generate a set of request likelihoods, each request likelihood representing a likelihood that the time-expiring inventory will receive a transaction request at each of a set of test price and test times to expiration. The system further trains a regression tree model based on a set of training data comprising each of the request likelihoods from the set and the test price and test time period to expiration used to generate the demand estimate that was used to generate the request likelihood.
1. A computer-executed method comprising:
receiving, at an online computing system associated with a booking accommodation application, a request for a price suggestion for a subject listing, the subject listing associated with a time-expiring inventory available to be booked by one of a plurality of clients of the online computing system;
generating a corpus of training data points associated with the time-expiring inventory, the training data points comprising a plurality of prices and a plurality of times until expiration of the time-expiring inventory;
generating an extrapolation set of times until expiration of the time-expiring inventory, the extrapolation set of times until expiration including times greater than those in the corpus of training data points;
inputting the extrapolation set of times until expiration into the regression tree model to determine a likelihood that the time-expiring inventory will receive a transaction request at each of the extrapolation set of times until expiration;
training a regression tree model to compress the corpus of training data points into a generalized model, the generalized model configured to output a price curve for the time-expiring inventory that is applicable regardless of which night the price curve is applied to, the compressing of the corpus of training data points enabling the regression tree model to avoid generating a separate price curve for each selectable night;
generating, using the regression tree model, the price curve for the time-expiring inventory for a given night, wherein the regression tree model represents a likelihood of receiving a transaction request for the time-expiring inventory within a duration until the time-expiring inventory expires;
determining, based on the generated price curve, a price suggestion for the time-expiring inventory and the given night; and
providing the determined price suggestion for display through a user interface of the booking accommodation application on a manager computing device.
2. The computer-executed method of claim 1 , further comprising:
receiving an indication of a change impacting the subject listing after training the regression tree model; and
adjusting the likelihood that the time-expiring inventory will receive a transaction request at each of the extrapolation set of times until expiration based on the indication of the change.
3. The computer-executed method of claim 1 , wherein determining the price suggestion comprises selecting a new price that maximizes a product of price for the time-expiring inventory multiplied by a likelihood that the time-expiring inventory will receive a transaction request for the given night.
4. The computer-executed method of claim 1 , further comprising:
receiving, from the manager computing device, a selection of a threshold likelihood of receiving a transaction request for the time-expiring inventory; and
determining, based on the generated price curve, a price suggestion that results in the threshold likelihood of receiving a transaction request for the given night.
5. The computer-executed method of claim 1 , wherein the price suggestion comprises a maximum price that results in at least a threshold likelihood of receiving a transaction request for the time-expiring inventory and the given night.
6. A non-transitory computer-readable medium at an online computing system encoding instructions executable by a processor to:
receive, at an online computing system associated with a booking accommodation application, a request for a price suggestion for a subject listing, the subject listing associated with a time-expiring inventory available to be booked by one of a plurality of clients of the online computing system;
generate a corpus of training data points associated with the time-expiring inventory, the training data points comprising a plurality of prices and a plurality of times until expiration of the time-expiring inventory;
generate an extrapolation set of times until expiration of the time-expiring inventory, the extrapolation set of times until expiration including times greater than those in the corpus of training data points;
input the extrapolation set of times until expiration into the regression tree model to determine a likelihood that the time-expiring inventory will receive a transaction request at each of the extrapolation set of times until expiration;
train a regression tree model to compress the corpus of training data points into a generalized model, the generalized model configured to output a price curve for the time-expiring inventory that is applicable regardless of which night the price curve is applied to, the compressing of the corpus of training data points enabling the regression tree model to avoid generating a separate price curve for each selectable night;
generate, using the regression tree model, the price curve for the time-expiring inventory for a given night, wherein the regression tree model represents a likelihood of receiving a transaction request for the time-expiring inventory within a duration until the time-expiring inventory expires;
determine, based on the generated price curve, a price suggestion for the time-expiring inventory and the given night; and
provide the determined price suggestion for display through a user interface of the booking accommodation application on a manager computing device.
7. The non-transitory computer-readable medium of claim 6 , the non-transitory computer readable medium encoding instructions executable by the processor to:
receive an indication of a change impacting the subject listing after training the regression tree model; and
adjust the likelihood that the time-expiring inventory will receive a transaction request at each of the extrapolation set of times until expiration based on the indication of the change.
8. The non-transitory computer-readable medium of claim 6 , wherein determining the price suggestion comprises selecting a new price that maximizes a product of price for the time-expiring inventory multiplied by a likelihood that the time-expiring inventory will receive a transaction request for the given night.
9. The non-transitory computer-readable medium of claim 6 , the non-transitory computer readable medium encoding instructions executable by the processor to:
receive, from the manager computing device, a selection of a threshold likelihood of receiving a transaction request for the time-expiring inventory; and
determine, based on the generated price curve, a price suggestion that results in the threshold likelihood of receiving a transaction request for the given night.
10. The non-transitory computer-readable medium of claim 6 , wherein the price suggestion comprises a maximum price that results in at least a threshold likelihood of receiving a transaction request for the time-expiring inventory and the given night.
11. A computer system comprising:
one or more computer processors for executing computer program instructions; and
a non-transitory computer-readable storage medium storing instructions executable by the one or more computer processors to perform steps comprising:
receiving, at an online computing system associated with a booking accommodation application, a request for a price suggestion for a subject listing, the subject listing associated with a time-expiring inventory available to be booked by one of a plurality of clients of the online computing system;
generating a corpus of training data points associated with the time-expiring inventory, the training data points comprising a plurality of prices and a plurality of times until expiration of the time-expiring inventory;
generating an extrapolation set of times until expiration of the time-expiring inventory, the extrapolation set of times until expiration including times greater than those in the corpus of training data points;
inputting the extrapolation set of times until expiration into the regression tree model to determine a likelihood that the time-expiring inventory will receive a transaction request at each of the extrapolation set of times until expiration;
training a regression tree model to compress the corpus of training data points into a generalized model, the generalized model configured to output a price curve for the time-expiring inventory that is applicable regardless of which night the price curve is applied to, the compressing of the corpus of training data points enabling the regression tree model to avoid generating a separate price curve for each selectable night;
generating, using the regression tree model, the price curve for the time-expiring inventory for a given night, wherein the regression tree model represents a likelihood of receiving a transaction request for the time-expiring inventory within a duration until the time-expiring inventory expires;
determining, based on the generated price curve, a price suggestion for the time-expiring inventory and the given night; and
providing the determined price suggestion for display through a user interface of the booking accommodation application on a manager computing device.
12. The computer system of claim 11 , wherein the steps further comprise:
receiving an indication of a change impacting the subject listing after training the regression tree model; and
adjusting the likelihood that the time-expiring inventory will receive a transaction request at each of the extrapolation set of times until expiration based on the indication of the change.
13. The computer system of claim 11 , wherein determining the price suggestion comprises selecting a new price that maximizes a product of price for the time-expiring inventory multiplied by a likelihood that the time-expiring inventory will receive a transaction request for the given night.
14. The computer system of claim 11 , wherein the steps further comprise:
receiving, from the manager computing device, a selection of a threshold likelihood of receiving a transaction request for the time-expiring inventory; and
determining, based on the generated price curve, a price suggestion that results in the threshold likelihood of receiving a transaction request for the given night.