Hotel income management method based on dynamic room price optimization

Abstract

The invention discloses a hotel income management method based on dynamic room price optimization, and relates to the technical field of hotel income management, and the method comprises the steps: obtaining room price data, carrying out the preprocessing of the room price data, and carrying out the feature extraction; building a housing price influence factor model based on the extracted features, and predicting the housing price according to the housing price influence factor model; and determining an optimal housing price interval according to the predicted housing price, optimizing the optimal housing price interval to obtain a final housing price, and realizing hotel income management based on the final housing price. According to the method, various machine learning algorithms are adopted, the house price influence factor model is constructed, the complex nonlinear relation between the house price and various factors can be captured, and the prediction result is more adaptive. The optimal housing price interval of the future seven days is generated through dynamic prediction of the model, the housing price can be adjusted in real time, and changes of market demands are responded. According to the dynamic optimization, the flexibility and the real-time performance of income management are remarkably improved, and the hotel is helped to keep advantages in competition.

Description

Technical Field

[0001] The invention relates to the technical field of hotel revenue management, in particular to a hotel revenue management method based on dynamic room rate optimization. Background Art

[0002] In the hotel industry, revenue management, as a key pricing strategy, has been widely used in recent years. As market competition intensifies, hotel managers increasingly rely on data-driven decision support systems to dynamically adjust room rates to maximize revenue. Traditional revenue management systems are mainly based on historical data analysis, and predict future room rates by evaluating factors such as market demand, competitor pricing, and seasonal fluctuations. However, these systems usually focus on static data, lack sensitivity to dynamic market changes, and cannot effectively respond to the impact of emergencies such as natural disasters, large-scale events, or policy changes.

[0003] Many existing systems rely on simple regression analysis of historical data and fail to fully consider the impact of dynamic market factors. For example, traditional models usually perform linear regression analysis based on past occupancy and room rate data. This method ignores the role of external variables such as macroeconomic fluctuations, regional activities, seasonal changes, and competitor price changes, resulting in prediction results that often lag behind actual market changes. This model that relies on static data is difficult to respond quickly to sudden changes in the market, which in turn affects revenue. In addition, the existing revenue management system lacks real-time dynamic optimization capabilities, resulting in insufficient flexibility in price adjustments. Many systems use fixed pricing strategies and are difficult to quickly adapt to the ever-changing market demand. During high-demand periods, traditional models may fail to increase room prices in a timely manner, resulting in the loss of revenue opportunities; while during low-demand periods, prices may not be adjusted in a timely manner to attract customers. This fixed pricing mechanism puts hotels at a disadvantage in market competition. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention proposes a hotel revenue management method based on dynamic room rate optimization.

[0005] Therefore, the problem to be solved by the present invention is that the existing revenue management system is often unable to respond quickly to market changes, resulting in delayed price adjustments.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, an embodiment of the present invention provides a hotel revenue management method based on dynamic room rate optimization, which comprises: Acquire housing price data, preprocess the housing price data, and perform feature extraction; Based on the extracted features, a housing price influencing factor model is constructed, and housing prices are predicted according to the housing price influencing factor model; An optimal price range is determined according to the predicted price, the optimal price range is optimized to obtain a final price, and hotel revenue management is achieved based on the final price.

[0007] As a preferred solution of the hotel revenue management method based on dynamic house price optimization described in the present invention, wherein: the house price data includes historical house price data and real-time market information; The historical housing price data includes standard prices, actual transaction prices, special discount prices, daily occupancy rates, average housing prices and available room revenue for different room types; The real-time market information includes search volume, booking volume, cancellation rate, tourist flow forecast data, weather, major event schedules and holidays.

[0008] As a preferred solution of the hotel revenue management method based on dynamic house price optimization described in the present invention, wherein: the house price influencing factor model includes a first algorithm, a second algorithm and a third algorithm; The first algorithm performs feature importance analysis to obtain feature importance results, and assigns weights according to the feature importance results; training the second algorithm according to the feature importance results and the weights; Input the output result of the second algorithm as an additional feature into the third algorithm, and perform weighted summation on the results of the first algorithm, the second algorithm and the third algorithm to obtain the result of the housing price influencing factor model; And the result of the housing price influencing factor model is used to update the feature importance result.

[0009] As a preferred solution of the hotel revenue management method based on dynamic room rate optimization of the present invention, wherein: the feature importance result includes the importance score of each feature; The importance score of each feature is calculated using the following formula: ;in, represents the weight of the i-th tree, is the information gain of feature f in the i-th tree, N is the total number of trees, is the feature importance score; The weights are assigned according to the feature importance results, as shown in the following formula: in, It is a feature The weight of is the temperature parameter, which controls the smoothness of the weight distribution. is an activation function used to normalize the numerical vector into a probability distribution.

[0010] As a preferred solution of the hotel revenue management method based on dynamic room rate optimization described in the present invention, the second algorithm includes an adaptive learning rate mechanism, as shown in the following formula: in, It is a feature The weight of is the temperature parameter, which controls the smoothness of the weight distribution. is an activation function used to normalize a numerical vector into a probability distribution.

[0011] As a preferred solution of the hotel revenue management method based on dynamic room rate optimization described in the present invention, the second algorithm includes an adaptive learning rate mechanism, as shown in the following formula: in, is the initial learning rate, is the adjustment amplitude, is the preset cycle; It is a feature The weight of is the time-varying learning rate; The third algorithm combines the attention mechanism and calculates the context vector to obtain the final output of the third algorithm. The attention mechanism is shown in the following formula: The final output of the third algorithm is as follows: in, , , is the weight matrix of the attention mechanism, is the cell state at the previous moment, is the bias vector of the attention mechanism, is the context vector obtained by weighted averaging of attention weights, is the output activation function, is the weight matrix of the output layer, is the bias vector of the output layer, is the output of the third algorithm, is the hidden state, is the prediction result of the second model.

[0012] As a preferred solution of the hotel revenue management method based on dynamic room rate optimization described in the present invention, wherein: using the result of the room rate influencing factor model to update the feature importance result includes updating the first algorithm in and importance scores; Update the first algorithm As shown below: The importance score is updated as follows: in, Represents the final prediction feature The partial derivative of reflects the characteristics The degree of impact on the final forecast, is a positive number that controls the magnitude of the update. represents the new weight of the i-th tree, is the price prediction value given by the i-th independent decision tree, is the final output of the housing price influencing factor model, is the control parameter of the penalty intensity, Score the new importance.

[0013] As a preferred solution of the hotel revenue management method based on dynamic room rate optimization described in the present invention, wherein: determining the optimal room rate range based on the predicted room rate includes using As the benchmark price, the confidence interval provided by the housing price influencing factor model is used to preliminarily determine the price range, and its upper and lower floating factors are calculated to finally obtain the dynamic price range, as shown in the following formula: in: ; ;in, and are the current and historical booking rates, and for current and historical competitor prices, is the impact factor of the emergency event, is the weight of the impact of the reservation rate on the price reduction, is the weight of the impact of competitor prices on price reduction, is the impact weight of the emergency on price reduction, is the weight of the impact of booking rate on price increase, is the weight of the impact of competitor prices on price increases, is the impact weight of the emergency on price increase, is the maximum adjustment range, and are the lower and upper adjustment factors of the price range, The dynamic price range.

[0014] As a preferred solution of the hotel revenue management method based on dynamic room rate optimization described in the present invention, the optimization of the optimal room rate range includes eliminating the sharp price fluctuations between adjacent dates and setting the maximum and minimum price limits for each room type, as shown in the following formula: in, To look back the number of days, is the current benchmark price weight, is the historical benchmark price weight, The minimum acceptable price set for the hotel, is the highest practicable price determined based on market research and historical data. is the final room price after optimization.

[0015] In a second aspect, an embodiment of the present invention provides a hotel revenue management system based on dynamic room rate optimization, which includes: A data collection and extraction module, used to obtain housing price data, pre-process the housing price data, and perform feature extraction; A model building module, used to build a housing price influencing factor model based on the extracted features, and predict housing prices based on the housing price influencing factor model; The room price determination module is used to determine the optimal room price range according to the predicted room price, optimize the optimal room price range to obtain the final room price, and realize the hotel's revenue management based on the final room price.

[0016] The beneficial effects of the present invention are as follows: a variety of machine learning algorithms are used to construct a model of factors affecting housing prices, which can capture the complex nonlinear relationship between housing prices and a variety of factors. Traditional methods can usually only handle static or linear factors and cannot effectively cope with dynamic market changes. The multi-model combination method of the present invention can effectively overcome this limitation and make the prediction results more adaptable. By generating the optimal housing price range for the next 7 days based on model-based dynamic prediction, the present invention can adjust housing prices in real time to respond to changes in market demand. Compared with the static pricing strategy in the prior art, this dynamic optimization significantly improves the flexibility and real-time nature of revenue management, helping hotels maintain their advantages in competition. The present invention effectively reduces price fluctuations between adjacent dates and avoids the instability of price strategies by optimizing and constraining the predicted price range. This robustness is rare in the prior art and can improve customers' stay experience, reduce churn, and enhance customer loyalty. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them: Figure 1 The flowchart of the hotel revenue management method based on dynamic room rate optimization is shown in FIG. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0019] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0020] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0021] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0022] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0023] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0024] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a hotel revenue management method based on dynamic room rate optimization, including: S1. Get historical housing price data and real-time market information First, use the hotel management system (PMS) to extract daily room rate data for the past 12 months. These data include standard prices, actual transaction prices, special discount prices, etc. for different room types. At the same time, record key performance indicators such as daily occupancy rate, average room rate (ADR) and available room revenue (RevPAR).

[0025] Secondly, by connecting to the API interface of major online travel platforms (OTA), we can obtain the room rate information of competitor hotels in real time, including the room rates of at least 5 competing hotels of the same level and region, as well as their price change trends in the next 7 days.

[0026] Third, collect market demand trend data, including search volume, booking volume, cancellation rate and other information obtained from OTA platforms, as well as tourist flow forecast data obtained from official channels such as tourism bureaus.

[0027] Fourth, integrate external factor data, such as weather forecasts, holiday schedules, major event schedules, etc. These factors have a significant impact on housing prices.

[0028] Finally, all collected data is stored in a specially designed dynamic housing price database. The database adopts a distributed storage architecture to ensure real-time update and fast access to data. The database also contains data cleaning and standardization modules to handle heterogeneous data from different sources and ensure data quality and consistency.

[0029] S2. Establish a model of factors affecting housing prices: Preprocess the large amount of data collected in S1. This includes data cleaning, outlier detection and processing, missing value filling, etc. The Z-score standardization method is used to normalize data of different dimensions to ensure that each factor is comparable in the model.

[0030] Furthermore, feature engineering techniques are used to extract and create meaningful features from the raw data. For example, dates can be converted into days of the week, whether it is a weekend, the number of days until the next holiday, etc.; weather data can be converted into discrete features such as temperature ranges and whether it is rainy; major events can be classified and coded, etc.

[0031] Furthermore, since housing prices are not only affected by static factors (the geographical location of the house, the age of construction, surrounding supporting facilities, etc.), but are also affected by many dynamic factors (changes in the economic environment, government policy adjustments, market supply and demand, etc.) over time, and there are complex interactions between these factors, the existing single prediction algorithm can only handle one aspect of them. Therefore, this embodiment proposes a housing price influencing factor model constructed by multiple machine learning algorithms.

[0032] For example, the model of factors affecting housing prices includes: First, the importance of each feature is evaluated using the improved random forest algorithm. The purpose of this step is to identify which factors have the greatest impact on housing prices. Specifically, multiple decision trees are constructed, and each tree uses a randomly selected subset of features. For each tree, the information gain brought by each feature is calculated. Then, the results of all trees are combined by weighted average to obtain the importance score of each feature. It can be expressed as follows: in, Represents the weight of the i-th tree. Initially, the weights of all trees are equal, but in subsequent iterations, the weights will be adjusted according to the prediction accuracy of each tree. is the information gain of feature f in the i-th tree, which measures the contribution of feature f to reducing the prediction error. N is the total number of trees, usually set between 100 and 1000. is the feature importance score.

[0033] It should be noted that the output of this step is a list of feature importances, for example: {"Room Type": 0.25, "Season": 0.2, "Weekend / Weekday": 0.15, "Holiday": 0.1, ...}.

[0034] Based on the feature importance results of random forest, feature screening and weight assignment are performed. The purpose of this step is to focus on the most relevant features and assign appropriate weights to different features. The softmax function is used to convert the importance scores into weights: in, is the temperature parameter, which controls the smoothness of the weight distribution. is an activation function used to normalize the numerical vector into a probability distribution. Values ​​of produce steeper distributions, highlighting important features; higher A value of 1 will produce a smoother distribution, giving equal weight to more features. For example, if = 0.1, we get the following weight distribution: {"Room Type": 0.4, "Season": 0.3, "Weekends / Weekdays": 0.2, "Holidays": 0.1, ...}.

[0035] Next, use the filtered features and assigned weights to train the XGBoost model and introduce an adaptive learning rate mechanism: in, is the initial learning rate (usually set to around 0.1), is the adjustment amplitude (e.g. 0.5), is the preset cycle; It is a feature The weight of is the time-varying learning rate.

[0036] It should be noted that the objective function of the traditional XGBoost model is as follows: in, is the loss function, which measures the difference between the predicted value and the true value. is a regularization term used to control the complexity of the model. is the true value of the i-th sample, is the predicted value of the ith sample, is the kth tree.

[0037] Its prediction function is as follows: in, is the feature vector of the i-th sample, is the set of all possible decision trees.

[0038] Adding the adaptive learning rate mechanism of the present invention changes the update steps of the XGBoost model, and we can obtain: in, is the predicted value of the i-th sample after t rounds of iterations, is the predicted value of the i-th sample after the t-1th iteration.

[0039] Preferably, the periodic change of the adaptive learning rate mechanism allows the XGBoost model to explore different learning rates multiple times during training, which helps to escape the local optimum, and combined with the feature weights, important features will get a larger learning rate, accelerating the model's learning of key features.

[0040] Furthermore, the prediction results of XGBoost are used as additional input features of LSTM to capture the time series characteristics of housing prices. Specifically, for each time step ,LSTM input includes not only the original features , which also includes XGBoost’s prediction at that time step In this way, the LSTM model can learn the nonlinear relationship captured by XGBoost while maintaining its sensitivity to time series patterns. The improved LSTM model can be expressed as follows: Input Gate: ; Forget Gate: ; Output Gate: ; Candidate cell states: ; Cell status update: ; Hide status updates: ; in, is the sigmoid function, is the hyperbolic tangent function, represents the Hadamard product (element-wise multiplication), is the hidden state at the previous moment, is the input at the current moment, is the prediction result of XGBoost, is the weight matrix, is the bias vector.

[0041] Furthermore, the output of LSTM is combined with the attention mechanism to calculate the context vector and obtain the final output of LSTM. The attention mechanism is shown in the following formula: The final output of the third algorithm is as follows: in, , , is the weight matrix of the attention mechanism, is the cell state at the previous moment, is the bias vector of the attention mechanism, is the context vector obtained by weighted averaging of attention weights, is the output activation function, is the weight matrix of the output layer, is the bias vector of the output layer, In hidden state.

[0042] Finally, the prediction results of random forest, XGBoost and LSTM are weighted and integrated to get the final predicted house price. , and use the final predicted house price To update the tree weights for random forest and feature importances for XGBoost: For Random Forest: This means that trees with more accurate predictions will receive higher weights. It is a tuning parameter that controls the magnitude of weight adjustment.

[0043] For XGBoost: in, Represents the final prediction feature The partial derivative of reflects the characteristics The degree of impact on the final forecast, is a small positive number that controls the magnitude of the update. is the control parameter of the penalty intensity.

[0044] S3: Based on the housing price influencing factor model established in S2, the optimal housing price range for the next 7 days is dynamically predicted using the time series analysis method to generate the upper and lower limits of daily housing prices.

[0045] First, prepare the forecast input data for the next 7 days. This includes time features (such as day of the week, weekend, holiday information), known booking data (such as current booking rate, number of days in advance), external factors (such as weather forecast, large event information, competitor published prices), and relevant economic indicator forecasts.

[0046] The house price influencing factor model constructed in step S2 (including random forest, XGBoost and LSTM) is used to make daily forecasts for the next 7 days. For each day, all feature data of that day are input into the model to obtain the benchmark house price forecast value and the forecast uncertainty range.

[0047] Next, determine the price range based on the model's predictions. As the benchmark price, the confidence interval provided by the model is used to preliminarily determine the price range. And the upper and lower floating factors are calculated to finally obtain the dynamic price range, which can be expressed as follows: in: ; ;in is the predicted price of the ensemble model at time t in step S2, and are the current and historical booking rates, and for current and historical competitor prices, is the impact factor of the emergency event, is the weight of the impact of the reservation rate on the price reduction, is the weight of the impact of competitor prices on price reduction, is the impact weight of the emergency on price reduction, is the weight of the impact of booking rate on price increase, is the weight of the impact of competitor prices on price increases, is the impact weight of the emergency on price increase, This is the maximum adjustment range.

[0048] Subsequently, the predicted price range is optimized and constrained. This includes ensuring that the price differences between different room types are reasonable, using moving average or exponential smoothing techniques to avoid sharp price fluctuations between adjacent dates, and setting maximum and minimum price limits for each room type. This is shown in the following formula: in For the integrated model in step S2 at time The predicted base price, To look back the number of days, is the current benchmark price weight, is the historical benchmark price weight, The minimum acceptable price set for the hotel, The highest practicable price determined based on market research and historical data.

[0049] final, The final forecast result will be The value is used as the benchmark for pricing in each sales channel. If the occupancy rate decreases or the cancellation rate increases during use, go back to S1 for optimization.

[0050] Furthermore, this embodiment also provides a hotel revenue management system based on dynamic room rate optimization, including: A data collection and extraction module, used to obtain housing price data, pre-process the housing price data, and perform feature extraction; A model building module, used to build a housing price influencing factor model based on the extracted features, and predict housing prices based on the housing price influencing factor model; The room price determination module is used to determine the optimal room price range according to the predicted room price, optimize the optimal room price range to obtain the final room price, and realize the hotel's revenue management based on the final room price.

[0051] This embodiment also provides a computer device, which is suitable for the hotel revenue management method based on dynamic room rate optimization, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the hotel revenue management method based on dynamic room rate optimization as proposed in the above embodiment.

[0052] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0053] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for hotel revenue management based on dynamic room rate optimization as proposed in the above embodiment is implemented.

[0054] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0055] Embodiment 2 is the second embodiment of the present invention. This embodiment provides a hotel revenue management method based on dynamic room rate optimization. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0056] This experiment selected a five-star hotel under an international hotel chain group located in the core business district of an economically developed city as the research object. The hotel has a total of 350 rooms, including standard rooms, deluxe rooms, executive rooms and suites. The experimental period is from March 1, 2024 to March 31, 2024. The reason for choosing this time period is that it spans ordinary working days, weekends and some holidays (Women's Day), which is highly representative.

[0057] Before the experiment began, the hotel management system (PMS) was upgraded to ensure that the system could support high-frequency data collection and real-time processing. By developing a dedicated API interface, data interoperability with major OTA platforms (Ctrip, Fliggy, booking, etc.) was achieved, and a real-time competitor price monitoring system was established. At the same time, a distributed storage architecture was deployed, using Redis as the cache layer and MongoDB as persistent storage to ensure the efficiency and reliability of data processing.

[0058] In the feature engineering stage, a total of 42 original features were collected and processed, including: room type features (area, orientation, facilities, etc.), time features (week, holidays, seasons, etc.), market features (competitor prices, market demand, etc.), weather features (temperature, precipitation, etc.) and event features (exhibitions, concerts, etc.). Through principal component analysis (PCA) and correlation analysis, 23 key features were finally selected for model training.

[0059] In the model training phase, a three-stage iterative optimization strategy was adopted: in the first stage, an improved random forest algorithm was used to evaluate feature importance, with the number of trees set to 500 and the minimum number of leaf node samples set to 50; in the second stage, the XGBoost model was trained, with the initial learning rate set to 0.08, the maximum depth set to 6, and the number of iterations set to 1000; in the third stage, the LSTM model was trained, using a bidirectional LSTM structure, a hidden layer dimension of 128, and a time step set to 14 days. The model training adopted the cross-validation method, with the data for the whole year of 2023 as the training set and the data from January to February 2024 as the validation set.

[0060] During the experiment, the model parameters are automatically updated at 2 a.m. every day, and the price forecast range for the next 7 days is generated. The system collects market data and fine-tunes prices every 4 hours to ensure that the price strategy can respond to market changes in a timely manner. At the same time, price fluctuation limits are set, and the price change between two consecutive days does not exceed 15% to maintain price stability and predictability. The following data are obtained: the price prediction accuracy statistics of different house types (March 2024) are shown in Table 1, and the importance ranking table of factors affecting house prices is shown in Table 2.

[0061] Table 1 Statistics of price prediction accuracy for different house types (March 2024) Room Type Sample size Mean absolute error (yuan) Root mean square error (yuan) Average relative error (%) Maximum error (yuan) Prediction accuracy (%) Price hit rate (%) Standard Room 930 45.6 58.2 4.8 125.3 92.5 88.7 Deluxe Room 620 62.3 79.1 5.2 168.4 91.8 87.2 Executive Room 465 89.7 112.5 5.9 235.6 90.2 85.9 suite 310 156.2 198.4 6.5 412.8 88.9 83.5 Table 2 Importance ranking of factors affecting housing prices Influencing factors Random Forest Importance Score XGBoost Importance Score Comprehensive score Weight coefficient Significance level Holiday Signs 0.892 0.875 0.884 0.156 0.001 Weekend sign 0.845 0.862 0.854 0.151 0.001 Competitor Price 0.823 0.848 0.836 0.148 0.001 Historical occupancy rate 0.812 0.825 0.819 0.145 0.001 Advance booking days 0.786 0.795 0.791 0.140 0.002 Weather conditions 0.745 0.752 0.749 0.132 0.002 Business Activities 0.712 0.728 0.720 0.128 0.003 As can be seen from the data in Table 1, the present invention shows high accuracy in price prediction for different room types. The prediction effect of standard room types is the best, with an average relative error of only 4.8% and a prediction accuracy of 92.5%. As the room type improves, the prediction difficulty gradually increases, but even for the highest-end suites, the prediction accuracy remains at a high level of 88.9%. This gradient prediction accuracy distribution is in line with the actual situation of the hotel industry, because the pricing of high-end room types is more complex.

[0062] Table 2 shows the key factors and their importance found in the experiment. Holiday signs and weekend signs received comprehensive scores of 0.884 and 0.854, respectively, ranking first and second in importance, indicating that the characteristics of the time dimension have the most significant impact on housing prices. Competitor prices (0.836) and historical occupancy rates (0.819) followed closely, reflecting the important influence of market competition and historical data on pricing decisions. The significance levels of all major influencing factors are below 0.005, confirming that there is a significant correlation between these factors and housing prices.

[0063] These data fully demonstrate the feasibility and superiority of the intelligent pricing system in practical applications. It not only solves the problems of traditional pricing methods, but also achieves significant improvements in revenue management. Through scientific pricing strategies and advanced algorithm models, the system maximizes hotel revenue while maintaining high customer satisfaction, providing a reliable technical solution for the intelligent transformation of the hotel industry.

[0064] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A hotel revenue management method based on dynamic room rate optimization, characterized by: include, Acquire housing price data, preprocess the housing price data, and perform feature extraction; Based on the extracted features, a housing price influencing factor model is constructed, and housing prices are predicted according to the housing price influencing factor model; An optimal price range is determined according to the predicted price, the optimal price range is optimized to obtain a final price, and hotel revenue management is achieved based on the final price.

2. The hotel revenue management method based on dynamic room rate optimization as claimed in claim 1, characterized in that: The housing price data includes historical housing price data and real-time market information; The historical housing price data includes standard prices, actual transaction prices, special discount prices, daily occupancy rates, average housing prices and available room revenue for different room types; The real-time market information includes search volume, booking volume, cancellation rate, tourist flow forecast data, weather, major event schedules and holidays.

3. The hotel revenue management method based on dynamic room rate optimization as claimed in claim 2, characterized in that: The housing price influencing factor model includes a first algorithm, a second algorithm and a third algorithm; The first algorithm performs feature importance analysis to obtain feature importance results, and assigns weights according to the feature importance results; training the second algorithm according to the feature importance results and the weights; Input the output result of the second algorithm as an additional feature into the third algorithm, and perform weighted summation on the results of the first algorithm, the second algorithm and the third algorithm to obtain the result of the housing price influencing factor model; And the result of the housing price influencing factor model is used to update the feature importance result.

4. The hotel revenue management method based on dynamic room rate optimization as claimed in claim 3, characterized in that: The feature importance result includes the importance score of each feature; The importance score of each feature is calculated using the following formula: ;in, represents the weight of the i-th tree, is the information gain of feature f in the i-th tree, N is the total number of trees, is the feature importance score; The weights are assigned according to the feature importance results, as shown in the following formula: in, It is a feature The weight of is the temperature parameter, which controls the smoothness of the weight distribution. is an activation function used to normalize the numerical vector into a probability distribution.

5. The hotel revenue management method based on dynamic room rate optimization as claimed in claim 4, characterized in that: The second algorithm includes an adaptive learning rate mechanism, as shown in the following formula: in, is the initial learning rate, is the adjustment amplitude, is the preset cycle; It is a feature The weight of is the time-varying learning rate; The third algorithm combines the attention mechanism and calculates the context vector to obtain the final output of the third algorithm. The attention mechanism is shown in the following formula: The final output of the third algorithm is as follows: in, , , is the weight matrix of the attention mechanism, is the cell state at the previous moment, is the bias vector of the attention mechanism, is the context vector obtained by weighted averaging of attention weights, is the output activation function, is the weight matrix of the output layer, is the bias vector of the output layer, is the output of the third algorithm, is the hidden state, is the prediction result of the second model.

6. The hotel revenue management method based on dynamic room rate optimization as claimed in claim 5, characterized in that: Using the result of the housing price influencing factor model to update the feature importance result includes updating the first algorithm and importance scores; Update the first algorithm As shown below: The importance score is updated as follows: in, Represents the final prediction feature The partial derivative of reflects the characteristics The degree of impact on the final forecast, is a positive number that controls the magnitude of the update. represents the new weight of the i-th tree, is the price prediction value given by the i-th independent decision tree, is the final output of the housing price influencing factor model, is the control parameter of the penalty intensity, Score the new importance.

7. The hotel revenue management method based on dynamic room rate optimization as claimed in claim 6, characterized in that: Determining the optimal price range based on the predicted price includes using As the benchmark price, the confidence interval provided by the housing price influencing factor model is used to preliminarily determine the price range, and its upper and lower floating factors are calculated to finally obtain the dynamic price range, as shown in the following formula: in: ; ;in, and are the current and historical booking rates, and for current and historical competitor prices, is the impact factor of the emergency event, is the weight of the impact of the reservation rate on the price reduction, is the weight of the impact of competitor prices on price reduction, is the impact weight of the emergency on price reduction, is the weight of the impact of booking rate on price increase, is the weight of the impact of competitor prices on price increases, is the impact weight of the emergency on price increase, is the maximum adjustment range, and are the lower and upper adjustment factors of the price range, The dynamic price range.

8. The hotel revenue management method based on dynamic room rate optimization as claimed in claim 7, characterized in that: Optimizing the optimal price range includes eliminating sharp price fluctuations between adjacent dates and setting maximum and minimum price limits for each room type, as shown in the following formula: in, To look back the number of days, is the current benchmark price weight, is the historical benchmark price weight, The minimum acceptable price set for the hotel, is the highest practicable price determined based on market research and historical data. is the final room price after optimization.

9. A hotel revenue management system based on dynamic room rate optimization, based on the hotel revenue management method based on dynamic room rate optimization according to any one of claims 1 to 8, characterized in that: include, A data collection and extraction module, used to obtain housing price data, pre-process the housing price data, and perform feature extraction; A model building module, used to build a housing price influencing factor model based on the extracted features, and predict housing prices based on the housing price influencing factor model; The room price determination module is used to determine the optimal room price range according to the predicted room price, optimize the optimal room price range to obtain the final room price, and realize the hotel's revenue management based on the final room price.

Images