CN109426887A - A kind of prediction technique and device of hotel inventory - Google Patents
A kind of prediction technique and device of hotel inventory Download PDFInfo
- Publication number
- CN109426887A CN109426887A CN201710770821.7A CN201710770821A CN109426887A CN 109426887 A CN109426887 A CN 109426887A CN 201710770821 A CN201710770821 A CN 201710770821A CN 109426887 A CN109426887 A CN 109426887A
- Authority
- CN
- China
- Prior art keywords
- house type
- predicted
- full
- state
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/12—Hotels or restaurants
Abstract
The application discloses the prediction technique and device of a kind of hotel inventory, this method comprises: determining the state of house type to be predicted according to the house type to be predicted of acquisition;The state of the house type to be predicted is predicted using corresponding prediction model according to the state of the house type to be predicted;According to the prediction result, the prediction quantity in stock of the house type to be predicted is obtained.Therefore, it is predicted using this method by state of the prediction model to house type to be predicted, with the quantity in stock of the determination house type to be predicted.Also, this method is the prediction result that the comparison of relevant information in state and prediction model using house type to be predicted obtains.It therefore, is more accurately, the case where on the one hand can not being subscribed to avoid the source of houses of closing, so that source can be sold house and bring considerable value income by improving using the quantity in stock that this method is got.On the other hand, the source of houses for having expired room is closed in time, can also be reduced full room list and be refused single rate, in short, can promote user experience using this method.
Description
Technical field
This application involves the technical fields of hotel inventory, and in particular to a kind of prediction technique and device of hotel inventory.
Background technique
It in online tourism industry, still, is booked rooms in most by the predetermined hotel guest room of platform of booking rooms on the net
On platform, the status information for the hotel's house type for the side of selling online that cannot timely update at all, user envisions certain for ordering certain date
The guest room in hotel is the state in full room by the hotel for booking rooms platform inquiry, does not receive to make a reservation for, still, probably due to letter
Breath update not in time, the house type in the hotel and non-full room state, really can receive it is scheduled, therefore, in this case, will
It makes troubles to the practical operation of user.
In addition, the status information of above-mentioned house type manually also can not quick response, and due to the variation of inventory's multidate information
Fastly, data accuracy is affected, and in addition the reason of hotel itself, which can not link, obtains inventory and lead to online that the side of selling can not be real
Apply the inventory for getting hotel's house type.
Therefore, it is necessary to the states that a kind of mode can more accurately determine house type, to offer convenience for the use of user.
Summary of the invention
The application provides the prediction technique of hotel inventory a kind of, to solve the above-mentioned problems in the prior art.
In addition the application provides the prediction meanss of hotel inventory a kind of.
The application provides the prediction technique of hotel inventory a kind of, this method comprises:
According to the house type to be predicted of acquisition, the state of house type to be predicted is determined;
The state of the house type to be predicted is carried out using corresponding prediction model according to the state of the house type to be predicted
Prediction;
According to the prediction result, the prediction quantity in stock of the house type to be predicted is obtained.
Optionally, the state according to the house type to be predicted, using corresponding prediction model, to the house type to be predicted
State predicted, comprising:
If the state of house type to be predicted is full room state, predicted using the room model that is full of in prediction model, root
It is predicted that result determines the predicted state of house type to be predicted;
If the state of house type to be predicted is non-full room state, room model is expired using the pass in prediction model and is predicted,
The predicted state of house type to be predicted is determined according to prediction result.
Optionally, described to be full of room model prediction using in prediction model, house type to be predicted is determined according to prediction result
Predicted state, comprising:
Using room model prediction is full of, judge whether to meet the trigger condition for opening house type;
If so, the predicted state of the house type to be predicted is revised as non-full room state;
If it is not, the predicted state of the house type to be predicted keeps original full room state.
Optionally, described using room model prediction is full of, judge whether to meet in the trigger condition for opening house type, it is described
Room model is full of to establish in the following ways:
The online house type for choosing preset quantity is as sampled data;
Determine the time of day of the house type in the sampled data;
Using the associated data using the house type in data as the input for being full of room model, by the true of the house type of the determination
Real state is as the output for being full of room model, to be full of the study and training of room model output and input as basic matching degree;
The trigger condition for opening house type for being full of room model is obtained according to the result of training and study.
Optionally, the triggering item for opening house type for being full of room model is obtained according to training study and result described
Part, comprising:
According to the result of training and study to the matching degree of the state of house type output and input, matching degree score value is obtained;
The matching degree score value is less than or equal to preset threshold as the trigger condition for opening house type.
Optionally, it is described using using the associated data of the house type in data as being full of in the input of room model, the pass
Join data include combination one or more kinds of in following information: the order situation of house type, the pricing information of house type, house type ground
Manage position.
Optionally, the study and training of the matching degree based on being full of the outputting and inputting of room model, comprising: adopt
Study and training with sorting algorithm to the matching degree output and input for being full of room model.
Optionally, after the online house type for choosing preset quantity is as sampled data, following operation is executed:
It is divided into training set data and test set data using data for described;
The time of day of house type in the determination sampled data, comprising: determine training set in the sampled data
The time of day of house type in data;
The associated data using using the house type in data is as the input for being full of room model, comprising: will use data
In training set data house type associated data as the input for being full of room model;
It is described according to training and study result obtain this be full of room model by house type open trigger condition after,
Execute following operation:
It is tested using the test set data in the sampled data room model is full of;
It is full of whether room model needs to optimize setting according to test result determination.
Optionally, the test set data using in data are tested room model is full of, comprising:
Obtain the time of day of house type in the test set data;
House type in the test set data is obtained into the state determined by model by the room model that is full of;
The time of day of house type is compared with the state determined by model;
It is described to be full of whether room model needs to optimize setting according to test result determination, comprising:
It is described when the time of day of house type ratio identical with the state determined by model is more than or equal to preset value
Room model is full of not need to optimize setting;
It is described to be full of room mould when the time of day of house type ratio identical with the state determined by model is less than preset value
Type needs optimize setting.
Optionally, described to be full of after room model needs to optimize setting, execute following operation:
Choose sampled data of the house type of preset quantity as optimization again online;
Determine the time of day of the house type in the sampled data;
Using the associated data using the house type in data as the input for being full of room model, by the true of the house type of the determination
Real state is as the output for being full of room model, to be full of the study and training of room model output and input as basic matching degree;
The trigger condition for opening house type for being full of room model is obtained according to the result of training and study;
Room model is full of using the model with the condition of setting out as after optimization.
Optionally, it before the house type to be predicted according to acquisition, the state for determining house type to be predicted, executes following
Operation:
Preset the detection time section of the state of house type;
The house type to be predicted according to acquisition, determines the state of house type to be predicted, comprising:
Often it is separated by the detection time section, the state of house type to be predicted is detected, by the house type to be predicted
Current state be set as the state of house type to be predicted.
Optionally, if the state of the house type to be predicted is non-full room state, Fang Mo is expired using the pass in prediction model
Type prediction, the predicted state of house type to be predicted is determined according to prediction result, comprising:
Using full room model prediction is closed, judge whether to meet the trigger condition for closing house type;
If so, the predicted state of house type to be predicted is revised as full room state;
If it is not, the predicted state of the house type to be predicted keeps original non-full room state.
Optionally, described using full room model prediction is closed, judge whether to meet the trigger condition for closing house type, the pass
Full room model is established in the following ways:
The online house type for choosing preset quantity is as sampled data;
Determine the time of day of the house type in the sampled data;
The associated data of the house type in data will be used as the input for closing full room model, by the true of the house type of the determination
Real state is as the output for closing full room model, to close the study and training of full room model output and input as basic matching degree;
Obtaining the pass according to the result of training and study expires the trigger condition for closing house type of room model.
Optionally, after the online house type for choosing preset quantity is as sampled data, following operation is executed:
It is divided into training set data and test set data using data for described;
The time of day of house type in the determination sampled data, comprising: determine training set in the sampled data
The time of day of house type in data;
It is described that the associated data of the house type in data will be used as the input for closing full room model, comprising: data will be used
In training set data house type associated data as the input for closing full room model;
It is described according to training and study result obtain the pass expire room model by house type opening trigger condition after,
Execute following operation:
It is tested using the test set data using in data full room model is closed;
Determine whether the full room model in the pass needs to optimize setting according to test result.
The application also provides the prediction meanss of hotel inventory a kind of, which includes:
Determination unit determines the state of house type to be predicted for the house type to be predicted according to acquisition;
Predicting unit, for the state according to the house type to be predicted, using corresponding prediction model, to the room to be predicted
The state of type is predicted;
Acquiring unit, for obtaining the prediction quantity in stock of the house type to be predicted according to the prediction result.
Optionally, the predicting unit includes:
It is full of room model unit, if the state for house type to be predicted is full room state, using opening in prediction model
Full room model is predicted, the predicted state of house type to be predicted is determined according to prediction result;
Full room model unit is closed, if the state for house type to be predicted is non-full room state, using in prediction model
It closes full room model to be predicted, the predicted state of house type to be predicted is determined according to prediction result.
Optionally, the room model unit that is full of includes:
Judgment sub-unit meets the trigger condition for opening house type for judging whether using room model prediction is full of;
Non-full room state subelement, for when the judging result of above-mentioned judgment sub-unit be when, by the room to be predicted
The predicted state of type is revised as non-full room state;
Full room state subelement, for when the judging result of above-mentioned judgment sub-unit is no, the house type to be predicted
Predicted state keeps original full room state.
Optionally, the judgment sub-unit includes:
Subelement is chosen, for choosing the house type of preset quantity online as sampled data;
Subelement is determined, for determining the time of day of the house type in the sampled data;
Training subelement, for using the associated data of the house type used in data as the input for being full of room model, by institute
The time of day of determining house type is stated as the output for being full of room model, is matched based on being full of the outputting and inputting of room model
The study and training of degree;
Trigger condition forms subelement, is full of the beating house type of room model for obtaining this according to the result of training and study
The trigger condition opened.
Optionally, the trigger condition formation subelement includes:
Matching degree score value obtains subelement, for the outputting and inputting to the state of house type according to the result of training and study
Matching degree, obtain matching degree score value;
Subelement is determined, for the matching degree score value to be less than or equal to preset threshold as the touching for opening house type
Clockwork spring part.
Optionally, the completely room model unit that closes includes:
Judgment sub-unit, for judging whether to meet the trigger condition for closing house type using full room model prediction is closed;
Full room state subelement, for when the judging result of the judgment sub-unit, which is, is, by house type to be predicted
Predicted state is revised as full room state;
Non-full room state subelement, for when the judging result of the judgment sub-unit be it is no when, the room to be predicted
The predicted state of type keeps original non-full room state.
Compared with prior art, the application has the following advantages:
The present invention provides the prediction technique of hotel inventory a kind of, method includes the following steps: according to the to be predicted of acquisition
House type, determine the state of house type to be predicted;According to the state of the house type to be predicted, using corresponding prediction model, to this
The state of house type to be predicted is predicted;According to the prediction result, the prediction quantity in stock of the house type to be predicted is obtained.
The application can be predicted using the above method by state of the prediction model to house type to be predicted, to determine
State the quantity in stock of house type to be predicted.Also, this method is relevant information in state and prediction model using house type to be predicted
Compare the prediction result obtained.Therefore, using the quantity in stock that this method is got be more accurately, on the one hand can be to avoid pass
The case where source of houses closed can not subscribe, so that source can be sold house and bring considerable value income by improving.On the other hand, it closes in time
The source of houses in room is expired, full room list can also be reduced and refuse single rate, in short, user experience can be promoted using this method.
Detailed description of the invention
Fig. 1 is the flow chart of the prediction technique for the hotel inventory that the application first embodiment provides.
Fig. 2 is the structural schematic diagram of the prediction meanss for the hotel inventory that the application second embodiment provides.
Specific embodiment
The application first embodiment provides the prediction technique of hotel inventory a kind of, and which is applied to for hotel's house type
The scene of the real-time update of state, using this method in the scene, it is ensured that the state of online house type is utmostly and very
Real state is identical, provides convenience to prepare to order the user of hotel guest room.
The application can be predicted using the above method by state of the prediction model to house type to be predicted, to determine
State the quantity in stock of house type to be predicted.Also, this method is relevant information in state and prediction model using house type to be predicted
Compare the prediction result obtained.Therefore, using the quantity in stock that this method is got be more accurately, on the one hand can be to avoid pass
The case where source of houses closed can not subscribe, so that source can be sold house and bring considerable value income by improving.On the other hand, it closes in time
The source of houses in room is expired, full room list can also be reduced and refuse single rate, in short, user experience can be promoted using this method.
It is described in detail and illustrates below by way of the method that specific embodiment provides the application first embodiment.
Fig. 1 is the flow chart of the prediction technique for the hotel inventory that the application first embodiment provides.Please refer to Fig. 1, this reality
Apply example offer method the following steps are included:
Step S101 determines the state of house type to be predicted according to the house type to be predicted of acquisition.
The step is to obtain the process of the current state of the house type to be predicted, generally by booking system or is booked rooms flat
The state of the house type to be predicted directly acquired on platform.And the not necessarily actual house type of current state obtained
State, for example, it is also possible to which there are following situations: the current state of the house type to be predicted of acquisition is full room state, and practical
It may be the case that the house type is non-full room state, the generation of the error may be since booking system or platform of booking rooms be not timely
Information caused by update is inconsistent.Therefore, based on the step, on the basis for the current state that the house type to be predicted determines
On further the virtual condition of the house type to be predicted is predicted, and then can determine the prediction inventory of house type to be predicted
Amount.
The current state institute that the house type is generally accompanied with when obtaining the current state of the house type to be predicted in the step is right
That answers moves in date and time, because according to the demand of user, when state of the most concerned house type of user is related with the date is moved in
Connection, therefore, the current state of the house type obtained herein are also and move between the time with corresponding relationship.
By taking hotel industry as an example, the current state of the house type to be predicted determined through the above steps first then can root
According to the difference of the house type current state to be predicted, forecast assessment is carried out according to corresponding prediction model respectively, then finally
The prediction quantity in stock of the house type to be predicted can be obtained according to prediction result.
It is the state of determining house type to be predicted above, after determining state, following steps can be executed.
It is right using corresponding prediction model according to the state of the house type to be predicted please continue to refer to Fig. 1, step S102
The state of the house type to be predicted is predicted.
The step is according to the state of the prediction house type, using corresponding prediction model, to the shape of the house type to be predicted
The process that state is predicted, it is, being carried out to the state of the house type to be predicted further pre- according to corresponding prediction model
The process of survey.
Specific prediction embodiment is as follows:
Following steps be for it is current be full two kinds of situations of room state and non-full room state respectively according to the pre- assessment of model
Estimate, be specifically described as follows:
The state according to the house type to be predicted, using corresponding prediction model, to the state of the house type to be predicted
It is predicted, including following implementation:
If the state of house type to be predicted is full room state, predicted using the room model that is full of in prediction model, root
It is predicted that result determines the predicted state of house type to be predicted.If the state of house type to be predicted is non-full room state, using prediction
Pass in model is expired room model and is predicted, the predicted state of house type to be predicted is determined according to prediction result.
The case where when being first full room state to the state of house type to be predicted, is described in detail and illustrates.
If the state of house type to be predicted is full room state, predicted using the room model that is full of in prediction model, root
It is predicted that result determines the predicted state of house type to be predicted.
The step introduction be the current state for the house type to be predicted determined to show full room state when the case where,
In such a case it is possible to then judge that this is to be predicted according to the result of prediction using room model prediction is full of for the house type
Whether the current display state of house type is close to prediction result, if it is, original display state is kept, if it is not, then will
The display status modifier of the house type is corresponding predicted state.
Specifically, described be full of room model prediction using in prediction model, house type to be predicted is determined according to prediction result
Predicted state, including following implementation: using room model prediction is full of, judge whether to meet the triggering item for opening house type
The predicted state of the house type to be predicted is if the determination result is YES revised as non-full room state by part;If judging result be it is no,
The predicted state of the house type to be predicted keeps original full room state.
It is, described be full of in room model includes trigger condition, when meeting the trigger condition, illustrate house type
State and the current state of display are different, and the predicted state by the house type to be predicted is needed to be revised as non-full room state.
But when being unsatisfactory for triggering, then the predicted state of the house type to be predicted keeps original full room state.
Need to illustrate when, it is described using being full of room model prediction, judge whether to meet the trigger condition for opening house type
When being specifically introduced, it should be introduced from the creation for being full of room model, the trigger condition can be described in detail.
Specifically, it is described using room model prediction is full of, judge whether to meet in the trigger condition for opening house type, it is described
Room model is full of to establish in the following ways:
The online house type for choosing preset quantity is as sampled data;Determine the true shape of the house type in the sampled data
State;Using the associated data using the house type in data as the input for being full of room model, by the true shape of the house type of the determination
State is as the output for being full of room model, to be full of the study and training of room model output and input as basic matching degree;According to
The result of training and study obtains the trigger condition for opening house type for being full of room model.
Wherein, it is described using using the associated data of the house type in data as being full of in the input of room model, the association
Data include combination one or more kinds of in following information: the order situation of house type, the pricing information of house type, house type geography
Position etc..
Specifically, the associated data can be some data for embodying the house type in the past period, which can be with
It is daily order volume (such as: 100 list/days), or order volume (such as: the 1000 list/moons) monthly etc., is also possible to
In past year, the case where which expires room rate.It or is the order situation or order of house type in the area
Price etc., the state that above-mentioned data can influence the house type to a certain extent is to expire the state in room or non-full room.
Above-mentioned data can be a series of parameter, for example, in the range of the parameter may include 200-300 group, at this
In range most effective immediate data reference can be provided to establish model.
Therefore, using above-mentioned associated data as the input parameter for being full of room model, accordingly, it is also necessary to described in determining
The time of day of house type in sampled data, as the output parameter for being full of room model.
And the side manually participated in is generally can be used into the process of the time of day of the house type in the determining sampled data
Formula obtains.It is, step chooses the house type of preset quantity as sampled data specifically by the way of online.
Specific embodiment is as follows:
The mode of artificial mark sample can be used, the method for using sampling randomly selects 1000 Real time displayings as full room
House type, and by the state of the modes such as phone or fax and hotel side's confirmation house type, and the state after confirming is the house type
Time of day.Although comparing cost time and efforts by the way of artificial mark sampling, it is obtained using which
The time of day of house type is that most accurately, therefore, from the aspect of accuracy, artificial mark belongs to by the way of most may be used
The mode leaned on.
Furthermore it is also possible to the time of day of house type is obtained using other modes, the time of day of any available house type
Mode belongs to the range of the embodiment of the present application protection.
Above-mentioned is to establish the process for being full of room model, and it includes the triggering item for opening house type that this, which is full of in room model,
Part then needs to modify the state that house type is currently shown according to the predicted state of house type to be predicted once meeting the trigger condition.
It should be noted that obtaining the touching for opening house type for being full of room model according to training study and result described
Clockwork spring part, comprising: according to the result of training and study to the matching degree of the state of house type output and input, obtain matching degree point
Value;The matching degree score value is less than or equal to preset threshold as the trigger condition for opening house type.
By inputting the matching degree with output, obtain corresponding matching degree score value, according to the score value judge whether to reach by
The trigger condition that the house type to be predicted is opened.
Finally, for using the associated data using the house type in data as the input for being full of room model, by the determination
House type time of day as the output for being full of room model, to be full of the outputting and inputting as basic matching degree of room model
It practises and trains;Trigger condition these steps for opening house type for being full of room model are obtained according to the result of training and study,
Wherein, the study and training of the matching degree based on being full of the outputting and inputting of room model, comprising: use sorting algorithm
Study and training to the matching degree output and input for being full of room model.Specifically, disaggregated model algorithm can be chosen, with
For machine forest model, training set training pattern is utilized.
In addition, can will be divided into two parts using data in order to which the room model that is full of is detected and optimized,
A part of sampled data is full of room model for training this, and the sampled data of another part can be used for testing this and be full of room mould
Whether type needs to optimize or improve.
Therefore, it for above situation, can be held after the online house type for choosing preset quantity is as sampled data
The following operation of row:
The sampled data is divided into training set data and test set data.
Correspondingly, the time of day of the house type in the determination sampled data, comprising: determine in the sampled data
The time of day of house type in training set data;It is described using using the associated data of the house type in data as being full of room model
Input, comprising: using the associated data using the house type of the training set data in data as the input for being full of room model.
When the sampled data includes training set data and test set data, correspondingly, described according to training and
The result of habit obtains this and is full of after the trigger condition for opening house type of room model, can execute following operation:
It is tested using the test set data using in data room model is full of;Institute is determined according to test result
It states and is full of whether room model needs to optimize setting.
Firstly, testing the step for progress to room model is full of to using the test set data using in data
Introduce, the step the following steps are included:
Obtain the time of day of house type in the test set data;House type in the test set data is full of by described
Room model obtains the state determined by model;The time of day of house type is compared with the state determined by model.Correspondingly,
After comparing, it is described according to test result determine described in be full of room model and whether need to optimize setting, including with
Lower two ways is distinguished:
It is one such are as follows: when the time of day of house type ratio identical with the state determined by model is more than or equal to
When preset value, the room model that is full of does not need to optimize setting;It is another are as follows: when house type time of day with it is true by model
When the identical ratio of fixed state is less than preset value, the room model needs that are full of optimize setting.
When identical large percentage, illustrating that this is full of the forecast assessment effect of room model is that comparison is accurately and stable,
Therefore, it does not need again to improve the model in this case.But when identical ratio is smaller, illustrate that this is full of room
The result of the prediction of model is always not accurate enough, therefore, in this case, it is further excellent to need to be full of this progress of room model
Change setting, to guarantee that the actual status information of house type can utmostly be embodied by being full of room model by this.
Therefore, if detected, when being full of room model to this and needing to optimize, correspondingly, the room model that is full of needs
After optimizing setting, following operation is executed: choosing sampled data of the house type of preset quantity as optimization again online;
Determine the time of day of the house type in the sampled data;Using using the associated data of the house type in data as being full of room model
Input, using the time of day of the house type of the determination as the output for being full of room model, to be full of the input of room model and defeated
It is out the study and training of basic matching degree;According to training and study result obtain this be full of room model by house type open
Trigger condition;Room model is full of using the model with the condition of setting out as after optimization.
It would know that, instructed again by data sampling again, and to the data of resampling through the above steps
Practice, and then is full of room model after being optimized.
Corresponding operating when full room state that be to the current state of house type to be predicted above be, and be non-for current state
When full room state, then room model is expired using the pass in prediction model and predicted, house type to be predicted is determined according to prediction result
Predicted state.
Illustrate first, when the current state is non-full room state, correspondingly, needing to be predicted using the full room model in pass
Assessment, and the time of day of house type can be determined according to prediction result.
And the foundation for closing full room model in above-mentioned steps to be full of room model foundation mode similar, also, root
It is predicted that result determine the time of day of house type with it is above-mentioned be full of room model when sentencing for time of day is determined according to prediction result
What the principle of disconnected method was also similar to.Therefore, the part that do not introduce in this step can refer to the foundation original for being full of room model
The introductory section of reason.
If expiring Fang Mo using the pass in prediction model specifically, the state of the house type to be predicted is non-full room state
Type prediction, the predicted state of house type to be predicted is determined according to prediction result, may include following implementation:
Using full room model prediction is closed, judge whether to meet the trigger condition for closing house type;If so, by room to be predicted
The predicted state of type is revised as full room state;If it is not, the predicted state of the house type to be predicted keeps original non-full room shape
State.
Wherein, described using full room model prediction is closed, judge whether to meet the trigger condition for closing house type, the pass is full
Room model is established in the following ways:
The online house type for choosing preset quantity is as sampled data;Determine the true shape of the house type in the sampled data
State;The associated data of the house type in data will be used as the input for closing full room model, by the true shape of the house type of the determination
State is as the output for closing full room model, to close the study and training of full room model output and input as basic matching degree;According to
The result of training and study, which obtains the pass, expires the trigger condition for closing house type of room model.
In addition, executing following operation after the online house type for choosing preset quantity is as sampled data:
It is divided into training set data and test set data using data for described;House type in the determination sampled data
Time of day, comprising: determine the time of day of the house type in the sampled data in training set data;It is described to use data
In house type associated data as the input for closing full room model, comprising: will be using the house type of the training set data in data
Associated data is as the input for closing full room model.
Correspondingly, obtaining the pass in the result according to training and study expires the triggering item for opening house type of room model
After part, following operation is executed:
It is tested using the test set data using in data full room model is closed;Institute is determined according to test result
State whether the full room model in pass needs to optimize setting.
The principle of the step is similar with above-mentioned steps, and above-mentioned steps are tested and assessed in advance to house type using being full of room model
Estimate, can be non-full room shape by full room status modifier by the state of the house type in the case where meeting the trigger condition of the model
State;Correspondingly, the step is to expire the clockwork spring out of room model when meeting the pass using full room model is closed to house type progress forecast assessment
It then can be full room state by non-full room status modifier by the state of the house type when the case where part.
It is above the state according to the house type to be predicted introduced, using corresponding prediction model, to the room to be predicted
The state of type is predicted, and the prediction model includes being full of room model and closing full room model to be predicted, is full of simultaneously
The trigger condition that corresponding house type is opened and house type is closed is separately included in the full room model of room model and pass, it can according to trigger condition
It is predicted with the state finally to the house type to be predicted, to obtain the state of prediction house type.
Please performance referring to Fig.1, step S103 obtains the prediction inventory of the house type to be predicted according to the prediction result
Amount.
The step is, according to the prediction result, to determine the prediction of the house type to be predicted on the basis of previous step
Quantity in stock.
The prediction quantity in stock is obtained according to above-mentioned prediction result, compared to the number directly obtained from platform of booking rooms
According to be predicted to being obtained on platform of booking rooms closer to true situation, therefore, after method provided by the present application
The current state of house type carries out further prediction and calculates, and is closest to very according to the state of the house type obtained after forecast assessment
Truth condition, therefore, the prediction quantity in stock acquired accordingly is also relatively true quantity in stock.
Also, method provided by the embodiments of the present application is training and study by the way of model, therefore, this method benefit
More accurate room state information is predicted with the prediction model of machine learning, the case where reducing the source of houses of closing can not subscribe improves
It can sell house and source and bring considerable value income.On the other hand, the source of houses for having expired room is closed in time, is reduced full room list and is refused list
Rate.
In short, this method provided using the application first embodiment is carried out in advance using the full room model of full model or pass is opened
It surveys, judges whether the current state of house type can embody its true state, and the time of day of house type can be replaced former
The current state come, to guarantee that the state of house type can be with real-time update.The accuracy rate of the state of real-time house type is promoted simultaneously.
In the above-described embodiment, the prediction technique of hotel inventory a kind of is provided, corresponding, the application second
Embodiment also provides the prediction meanss of hotel inventory a kind of.Fig. 2 is please referred to, a kind of wine provided for the application second embodiment
The schematic diagram of the embodiment of the prediction meanss of shop inventory.Since Installation practice is substantially similar to embodiment of the method, so description
Must be fairly simple, the relevent part can refer to the partial explaination of embodiments of method.Installation practice described below is only to show
Meaning property.
A kind of prediction meanss of the hotel inventory of the present embodiment, comprising:
Determination unit 201 determines the state of house type to be predicted for the house type to be predicted according to acquisition;
Predicting unit 202, for the state according to the house type to be predicted, using corresponding prediction model, to this wait for it is pre-
The state of ell type is predicted;
Acquiring unit 203, for obtaining the prediction quantity in stock of the house type to be predicted according to the prediction result.
Optionally, the predicting unit includes:
It is full of room model unit, if the state for house type to be predicted is full room state, using opening in prediction model
Full room model is predicted, the predicted state of house type to be predicted is determined according to prediction result;
Full room model unit is closed, if the state for house type to be predicted is non-full room state, using in prediction model
It closes full room model to be predicted, the predicted state of house type to be predicted is determined according to prediction result.
Optionally, the room model unit that is full of includes:
Judgment sub-unit meets the trigger condition for opening house type for judging whether using room model prediction is full of;
Non-full room state subelement, for when the judging result of above-mentioned judgment sub-unit be when, by the room to be predicted
The predicted state of type is revised as non-full room state;
Full room state subelement, for when the judging result of above-mentioned judgment sub-unit is no, the house type to be predicted
Predicted state keeps original full room state.
Optionally, the judgment sub-unit includes:
Subelement is chosen, for choosing the house type of preset quantity online as sampled data;
Subelement is determined, for determining the time of day of the house type in the sampled data;
Training subelement, for using the associated data of the house type used in data as the input for being full of room model, by institute
The time of day of determining house type is stated as the output for being full of room model, is matched based on being full of the outputting and inputting of room model
The study and training of degree;
Trigger condition forms subelement, is full of the beating house type of room model for obtaining this according to the result of training and study
The trigger condition opened.
Optionally, the trigger condition formation subelement includes:
Matching degree score value obtains subelement, for the outputting and inputting to the state of house type according to the result of training and study
Matching degree, obtain matching degree score value;
Subelement is determined, for the matching degree score value to be less than or equal to preset threshold as the touching for opening house type
Clockwork spring part.
Optionally, the completely room model unit that closes includes:
Judgment sub-unit, for judging whether to meet the trigger condition for closing house type using full room model prediction is closed;
Full room state subelement, for when the judging result of the judgment sub-unit, which is, is, by house type to be predicted
Predicted state is revised as full room state;
Non-full room state subelement, for when the judging result of the judgment sub-unit be it is no when, the room to be predicted
The predicted state of type keeps original non-full room state.
More accurate room state letter is predicted using the prediction model that device provided in this embodiment can use machine learning
The case where breath, reducing the source of houses of closing can not subscribe, source can be sold house and bring considerable value income by improving.On the other hand, and
When close and expire the source of houses in room, reducing, which expires room list, refuses single rate.
Although the application is disclosed as above with preferred embodiment, it is not for limiting the application, any this field skill
Art personnel are not departing from spirit and scope, can make possible variation and modification, therefore the guarantor of the application
Shield range should be subject to the range that the claim of this application defined.
In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net
Network interface and memory.Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM)
And/or the forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable Jie
The example of matter.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include non-temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
It will be understood by those skilled in the art that embodiments herein can provide as method, system or computer program product.
Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the application
Form.It is deposited moreover, the application can be used to can be used in the computer that one or more wherein includes computer usable program code
The shape for the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Formula.
Claims (15)
1. a kind of prediction technique of hotel inventory characterized by comprising
According to the house type to be predicted of acquisition, the state of house type to be predicted is determined;
The state of the house type to be predicted is predicted using corresponding prediction model according to the state of the house type to be predicted;
According to the prediction result, the prediction quantity in stock of the house type to be predicted is obtained.
2. the prediction technique of hotel inventory according to claim 1, which is characterized in that described according to the house type to be predicted
State the state of the house type to be predicted is predicted using corresponding prediction model, comprising:
If the state of house type to be predicted is full room state, predicted using the room model that is full of in prediction model, according to pre-
Survey the predicted state that result determines house type to be predicted;
If the state of house type to be predicted is non-full room state, room model is expired using the pass in prediction model and is predicted, according to
Prediction result determines the predicted state of house type to be predicted.
3. the prediction technique of hotel inventory according to claim 2, which is characterized in that described using opening in prediction model
Full room model prediction, the predicted state of house type to be predicted is determined according to prediction result, comprising:
Using room model prediction is full of, judge whether to meet the trigger condition for opening house type;
If so, the predicted state of the house type to be predicted is revised as non-full room state;
If it is not, the predicted state of the house type to be predicted keeps original full room state.
4. the prediction technique of hotel inventory according to claim 3, which is characterized in that described pre- using room model is full of
It surveys, judges whether to meet in the trigger condition for opening house type, the room model that is full of is established in the following ways:
The online house type for choosing preset quantity is as sampled data;
Determine the time of day of the house type in the sampled data;
Using the associated data using the house type in data as the input for being full of room model, by the true shape of the house type of the determination
State is as the output for being full of room model, to be full of the study and training of room model output and input as basic matching degree;
The trigger condition for opening house type for being full of room model is obtained according to the result of training and study.
5. the prediction technique of hotel inventory according to claim 4, which is characterized in that learn and tie according to training described
Fruit obtains the trigger condition for opening house type for being full of room model, comprising:
According to the result of training and study to the matching degree of the state of house type output and input, matching degree score value is obtained;
The matching degree score value is less than or equal to preset threshold as the trigger condition for opening house type.
6. the prediction technique of hotel inventory according to claim 4, which is characterized in that described to use the house type in data
Associated data as being full of in the input of room model, the associated data includes group one or more kinds of in following information
Close: the order situation of house type, the pricing information of house type, house type geographical location.
7. the prediction technique of hotel inventory according to claim 4, which is characterized in that the input to be full of room model
With the study and training of the matching degree based on output, comprising: using sorting algorithm to being full of outputting and inputting for room model
The study and training of matching degree.
8. the prediction technique of hotel inventory according to claim 4, which is characterized in that in the online selection preset quantity
House type as sampled data after, execute following operation:
It is divided into training set data and test set data using data for described;
The time of day of house type in the determination sampled data, comprising: determine training set data in the sampled data
In house type time of day;
The associated data using using the house type in data is as the input for being full of room model, comprising: will be using in data
The associated data of the house type of training set data is as the input for being full of room model;
It is described according to training and study result obtain this be full of room model by house type open trigger condition after, execute
It operates below:
It is tested using the test set data in the sampled data room model is full of;
It is full of whether room model needs to optimize setting according to test result determination.
9. the prediction technique of hotel inventory according to claim 8, which is characterized in that the test set using in data
Data are tested room model is full of, comprising:
Obtain the time of day of house type in the test set data;
House type in the test set data is obtained into the state determined by model by the room model that is full of;
The time of day of house type is compared with the state determined by model;
It is described to be full of whether room model needs to optimize setting according to test result determination, comprising:
It is described to be full of when the time of day of house type ratio identical with the state determined by model is more than or equal to preset value
Room model does not need to optimize setting;
When the time of day of house type ratio identical with the state determined by model is less than preset value, the room model that is full of is needed
Optimize setting.
10. the prediction technique of hotel inventory according to claim 9, which is characterized in that it is described be full of room model need into
After row optimal setting, following operation is executed:
Choose sampled data of the house type of preset quantity as optimization again online;
Determine the time of day of the house type in the sampled data;
Using the associated data using the house type in data as the input for being full of room model, by the true shape of the house type of the determination
State is as the output for being full of room model, to be full of the study and training of room model output and input as basic matching degree;
The trigger condition for opening house type for being full of room model is obtained according to the result of training and study;
Room model is full of using the model with the condition of setting out as after optimization.
11. the prediction technique of hotel inventory according to claim 1, which is characterized in that it is described according to acquisition to pre-
The house type of survey before the state for determining house type to be predicted, executes following operation:
Preset the detection time section of the state of house type;
The house type to be predicted according to acquisition, determines the state of house type to be predicted, comprising:
Often it is separated by the detection time section, the state of house type to be predicted is detected, by works as the house type to be predicted
Preceding state is set as the state of house type to be predicted.
12. the prediction technique of hotel inventory according to claim 2, which is characterized in that if the shape of the house type to be predicted
State is non-full room state, then expires room model prediction using the pass in prediction model, determine house type to be predicted according to prediction result
Predicted state, comprising:
Using full room model prediction is closed, judge whether to meet the trigger condition for closing house type;
If so, the predicted state of house type to be predicted is revised as full room state;
If it is not, the predicted state of the house type to be predicted keeps original non-full room state.
13. the prediction technique of hotel inventory according to claim 12, which is characterized in that described pre- using full room model is closed
It surveys, judges whether to meet the trigger condition for closing house type, the completely room model that closes is established in the following ways:
The online house type for choosing preset quantity is as sampled data;
Determine the time of day of the house type in the sampled data;
The associated data of the house type in data will be used as the input for closing full room model, by the true shape of the house type of the determination
State is as the output for closing full room model, to close the study and training of full room model output and input as basic matching degree;
Obtaining the pass according to the result of training and study expires the trigger condition for closing house type of room model.
14. the prediction technique of hotel inventory according to claim 13, which is characterized in that in the online selection present count
After the house type of amount is as sampled data, following operation is executed:
It is divided into training set data and test set data using data for described;
The time of day of house type in the determination sampled data, comprising: determine training set data in the sampled data
In house type time of day;
It is described that the associated data of the house type in data will be used as the input for closing full room model, comprising: will be using in data
The associated data of the house type of training set data is as the input for closing full room model;
It is described according to training and study result obtain the pass expire room model by house type open trigger condition after, execution
It operates below:
It is tested using the test set data using in data full room model is closed;
Determine whether the full room model in the pass needs to optimize setting according to test result.
15. a kind of prediction meanss of hotel inventory characterized by comprising
Determination unit determines the state of house type to be predicted for the house type to be predicted according to acquisition;
Predicting unit, for the state according to the house type to be predicted, using corresponding prediction model, to the house type to be predicted
State is predicted;
Acquiring unit, for obtaining the prediction quantity in stock of the house type to be predicted according to the prediction result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710770821.7A CN109426887A (en) | 2017-08-31 | 2017-08-31 | A kind of prediction technique and device of hotel inventory |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710770821.7A CN109426887A (en) | 2017-08-31 | 2017-08-31 | A kind of prediction technique and device of hotel inventory |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109426887A true CN109426887A (en) | 2019-03-05 |
Family
ID=65504683
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710770821.7A Pending CN109426887A (en) | 2017-08-31 | 2017-08-31 | A kind of prediction technique and device of hotel inventory |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109426887A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110950195A (en) * | 2019-11-29 | 2020-04-03 | 东软睿驰汽车技术(沈阳)有限公司 | Elevator optimization control method, device and equipment |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006285815A (en) * | 2005-04-04 | 2006-10-19 | Higashi Nihon Medicom Kk | Ordering prediction type stock management system by individual patient, ordering prediction type stock management program and recording medium therefor |
CN104933465A (en) * | 2015-05-12 | 2015-09-23 | 携程计算机技术(上海)有限公司 | Data interconnecting method and system on hotel management platform |
CN106780173A (en) * | 2016-12-01 | 2017-05-31 | 携程计算机技术(上海)有限公司 | OTA hotels inventory management method and system |
-
2017
- 2017-08-31 CN CN201710770821.7A patent/CN109426887A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006285815A (en) * | 2005-04-04 | 2006-10-19 | Higashi Nihon Medicom Kk | Ordering prediction type stock management system by individual patient, ordering prediction type stock management program and recording medium therefor |
CN104933465A (en) * | 2015-05-12 | 2015-09-23 | 携程计算机技术(上海)有限公司 | Data interconnecting method and system on hotel management platform |
CN106780173A (en) * | 2016-12-01 | 2017-05-31 | 携程计算机技术(上海)有限公司 | OTA hotels inventory management method and system |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110950195A (en) * | 2019-11-29 | 2020-04-03 | 东软睿驰汽车技术(沈阳)有限公司 | Elevator optimization control method, device and equipment |
CN110950195B (en) * | 2019-11-29 | 2022-03-15 | 东软睿驰汽车技术(沈阳)有限公司 | Elevator optimization control method, device and equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Anvari et al. | Automated Box–Jenkins forecasting tool with an application for passenger demand in urban rail systems | |
CN108345958A (en) | A kind of order goes out to eat time prediction model construction, prediction technique, model and device | |
US20120166616A1 (en) | System and method for energy performance management | |
US11816620B2 (en) | Plan modeling visualization | |
US20120072357A1 (en) | Method and system for predicting property values within discrete finite market elements | |
CN108550090A (en) | A kind of processing method and system of determining source of houses pricing information | |
US20170154287A1 (en) | Waste management simulation system and method | |
Hui et al. | Analysis of the viability of an urban renewal project under a risk-based option pricing framework | |
Azimi et al. | Applying basic control theory principles to project control: Case study of off-site construction shops | |
JP2019028871A (en) | Project management support device, project management support method and program | |
JP2009104408A (en) | Integrated demand forecasting apparatus, integrated demand forecasting method and integrated demand forecasting program | |
CN109615414A (en) | House property predictor method, device and storage medium | |
JP5551806B2 (en) | Integrated demand prediction apparatus, integrated demand prediction method, and integrated demand prediction program | |
Jin et al. | Exploring spatiotemporal dynamics in a housing market using the spatial vector autoregressive Lasso: A case study of Seoul, Korea | |
US20240078607A1 (en) | Digital system for controlled boost of expert forecasts in complex prediction environments and corresponding method thereof | |
CN109426887A (en) | A kind of prediction technique and device of hotel inventory | |
JPH09233700A (en) | Method of evaluating reliability on estimation of day maximum demand power | |
US20230306515A1 (en) | Systems and Computer-Implemented Methods for Capital Management | |
Ahlers et al. | Supporting Municipal Greenhouse Gas (GHG) emission inventories using business process modeling: a case study of Trondheim Municipality | |
US20150006342A1 (en) | Generating a Simulated Invoice | |
JP2015103246A (en) | Price determination method, system, and computer program (robust solution method of price determination for product and service) | |
US20220292550A1 (en) | Apparatuses, Systems, and Methods for Providing an Automated Valuation Model for Commercial Real Estate | |
Nair et al. | Indian Stock Market Forecasting using Prophet Model | |
Hwang | Essential contents for software development process and software quality education | |
WO2014162602A1 (en) | Index-setting assistance device and index-setting assistance method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190305 |