Incremental housing vacancy rate prediction method and device, medium and equipment

Abstract

The invention discloses an incremental housing vacancy rate prediction method and device, a medium and equipment. The method comprises the steps of obtaining electric power data used for housing vacancy rate prediction in a target area; constructing an incremental housing vacancy rate prediction model; and taking the power data as the input of the incremental housing vacancy rate prediction model to predict the incremental housing vacancy rate in a future preset period. The method can achieve the dynamic modeling and precise prediction of the increment housing vacancy rate, and improves the scientificity and timeliness of the supply and demand regulation of the urban housing.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of big data analysis, and particularly relates to an incremental housing vacancy rate prediction method. BACKGROUND

[0002] With the acceleration of urbanization and the development of the real estate market, the housing vacancy rate gradually becomes an important indicator for measuring the balance of urban housing supply and demand, the health of the market, and potential financial risks. In the prior art, the calculation of the incremental housing vacancy rate mainly relies on questionnaire surveys, on-site visits, and property data aggregation, which has problems such as data lag, insufficient precision, and high cost.

[0003] In recent years, with the popularity of smart electricity meters and smart water meters, it has become a trend to indirectly calculate housing usage by using energy consumption data such as electricity and water. However, there is currently a lack of a method that can dynamically reflect the trend of the incremental housing vacancy rate and has predictive ability, making it difficult to meet the needs of urban management and policy making. SUMMARY

[0004] In view of the deficiencies in the prior art, the purpose of the present application is to provide an incremental housing vacancy rate prediction method, device, medium, and equipment, which aims to accurately predict the incremental housing vacancy rate.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions: An incremental housing vacancy rate prediction method based on time series analysis, the prediction method comprising: obtaining electricity data for housing vacancy rate prediction in a target area; constructing an incremental housing vacancy rate prediction model; using the electricity data as the input of the incremental housing vacancy rate prediction model to predict the incremental housing vacancy rate in a future preset period.

[0006] Optionally, the electricity data for housing vacancy rate prediction in the target area comprises: obtaining monthly electricity data of incremental housing of urban residents in the target area, fitting the monthly electricity data according to a Gaussian mixture model to obtain a housing vacancy state electricity threshold; and calculating the current incremental housing vacancy rate based on the housing vacancy state electricity threshold.

[0007] Optionally, the construction of the incremental housing vacancy rate prediction model comprises: defining a survival analysis framework; under the survival analysis framework, constructing a state space model and describing the dynamic evolution process of the vacancy rate to associate unobserved survival parameters with observed data; based on the state space model, estimating the dynamic changes of the survival parameters in the state space model by a Markov chain Monte Carlo algorithm; and based on the model parameter estimation result, constructing the incremental housing vacancy rate prediction model in combination with the survival analysis framework.

[0008] Optionally, the joint probability density function of the state space model is constructed and the model parameters are estimated by a Markov chain Monte Carlo algorithm, comprising: determining the parameter prior distribution; updating the state variable from the posterior distribution of the state variable by the Metropolis-Hastings algorithm, updating the variance by the Gibbs sampling, and obtaining the estimation result of the state space model parameters.

[0009] The incremental housing vacancy rate prediction model is represented as:

[0010] Wherein, represents the prediction value of the entity at time ; corresponding variable at future time ; represents the change rate of the entity at time and ; represents the current time step; is an exponential function, representing a process decaying over time.

[0011] The application also provides an incremental housing vacancy rate prediction device based on time series analysis, comprising: an acquisition module configured to acquire power data for housing vacancy rate prediction in a target area; a model construction module configured to construct an incremental housing vacancy rate prediction model; a prediction module configured to take the power data as input of the incremental housing vacancy rate prediction model to predict the incremental housing vacancy rate in a future preset period.

[0012] The application also provides a storage medium comprising instructions which, when executed on a computer, cause the computer to perform the method of any one of the preceding.

[0013] The application also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of the preceding when executing the program.

[0014] Compared with the prior art, the application has the following beneficial effects: The application breaks through the limitation of traditional survey method data lag, high cost and low precision by introducing a survival analysis framework, combining a state space model to construct a dynamic risk function, and using power data to model and predict the incremental housing vacancy rate. Secondly, the application estimates the state space model parameters by Markov Chain Monte Carlo algorithm, which can effectively describe the nonlinear trend of the vacancy rate changing over time, dynamically reflect the market housing de-liquefaction speed, and realize the fine prediction of the future vacancy rate, providing high timeliness, high explanatory power and high adaptability of quantitative decision support for urban housing supply and demand regulation, land transfer rhythm management and real estate policy making.

[0015] Power data BRIEF DESCRIPTION OF DRAWINGS Figure 1 is a flowchart of a kind of incremental housing vacancy rate prediction method based on time series analysis provided in an embodiment of the application; Figure 2 is a density map of monthly electricity consumption of urban residents provided in another embodiment of the application; Figure 3 is a curve graph of incremental housing vacancy rate in a certain place in 2021 and 2022 provided in another embodiment of the application; Figure 4 is a structure schematic diagram of a kind of incremental housing vacancy rate prediction device based on time series analysis provided in another embodiment of the application. DETAILED DESCRIPTION

[0016] The specific embodiments of the application will be described in detail below with reference to the accompanying drawings. Although specific embodiments of the application are shown in the drawings, it should be understood that the application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the application and to fully convey the scope of the application to those skilled in the art.

[0017] It should be noted that certain terms are used in the specification and claims to refer to particular components. Those skilled in the art will understand that the same component can be referred to by different names. The specification and claims of this specification do not distinguish components based on the difference in name, but rather on the difference in function. As used throughout the specification and claims, "comprising" or "including" is an open term, which should be interpreted as "including but not limited to". The subsequent description of the specification is a preferred embodiment of the application, which is intended to illustrate the general principles of the specification, and is not intended to limit the scope of the application. The scope of protection of the application is defined by the appended claims.

[0018] For the convenience of understanding the embodiments of the present application, further explanation and description will be made below with specific embodiments as examples in combination with the accompanying drawings, and each drawing does not constitute a limitation to the embodiments of the present application.

[0019] Figure 1 is a flowchart of a time series analysis-based incremental housing vacancy rate prediction method provided by an embodiment of the present application, as shown in Figure 1 , the method comprises the following steps: S100: obtaining power data in a target area for housing vacancy rate prediction; S200: constructing an incremental housing vacancy rate prediction model and training; S300: taking the power data as the input of the trained incremental housing vacancy rate prediction model to predict the incremental housing vacancy rate in a future preset period (for example, three years).

[0020] In another exemplary embodiment, in step S100, the power data in the target area for housing vacancy rate prediction is obtained, comprising the following steps: S101: obtaining monthly electricity consumption data of incremental housing of urban residents in the target area, fitting the monthly electricity consumption data according to a Gaussian mixture model to obtain a housing vacancy state electricity threshold; In this step, since there are significant differences in electricity consumption in different months, in order to ensure the reliability of the results, the present application selects representative March and October electricity consumption data of urban residents in each county of Fujian Province.

[0021] Figure 2 The density map of the monthly electricity consumption of urban residents after logarithmic transformation (let x=log(y+0.01), where y represents the monthly electricity consumption) is shown, which is similar to the Gaussian mixture distribution. From Figure 2 , it can be found that the electricity consumption of different types of houses shows obvious clustering effect. Specifically, there are obvious peak clustering of various electricity data such as vacant houses and low electricity houses, normal electricity houses and high electricity houses. Therefore, the present application considers using a Gaussian mixture model to find the segmentation point for judging the vacancy of houses.

[0022] The Gaussian mixture model is a probability model, which is a mixture distribution composed of multiple Gaussian distributions with different weights. From , the probability density function of a one-dimensional Gaussian mixture model composed of gaussian distributions and data from the th gaussian distribution is as follows:

[0023]

[0024] wherein, represents the probability density function value of the data point ; represents the number of components of the Gaussian distribution in the mixture model; represents the exponential function, is a discrete variable, indicates that the data comes from the th Gaussian sub-distribution, when , , otherwise ; represents the probability density function of the th Gaussian distribution; and are the mean and variance of the th Gaussian distribution, respectively; represents the posterior probability of the data point belonging to the th Gaussian distribution component; represents the prior probability of the th Gaussian distribution component; represents the probability density function of the th Gaussian distribution when the data point ; represents the mixing weight of the th Gaussian distribution component; represents the mixing weight of the th Gaussian distribution component; represents the probability density of the th Gaussian distribution component;

[0025] For this model, the model parameters need to be estimated based on the monthly electricity consumption data after logarithmic transformation; and further obtain the electricity threshold for determining the vacancy of the house. The parameters of the Gaussian mixture model can be estimated by the EM algorithm.

[0026] Specifically, the core idea of the EM algorithm is to maximize the likelihood function by iteratively replacing the E step and the M step: in the E step, based on the current estimate of the parameter, the expectation of the log-likelihood function of the complete data given the known data is calculated; in the M step, the parameter estimate value is solved by maximizing the E step expectation. In the Gaussian mixture model, let where is the known data, and is the missing data, so the Gaussian mixture model parameters can be obtained by the EM algorithm.

[0027] Further, the iteration process of the EM algorithm is as follows: 1. Step E, calculate the expectation of the log-likelihood function of the complete data given the known data, specifically including: (1) The formula of the log-likelihood function of the complete data is:

[0028] wherein, and represent the mean and variance of the first Gaussian distribution, respectively; (2) The expectation of the log-likelihood function of the complete data given the known data is:

[0029] wherein, let , then:

[0030]

[0031] wherein, represents a constant; 2. Step M, maximize the expectation of Step E by solving the extreme value to obtain the parameter estimate, specifically, by setting the first-order derivative of to zero, the parameter estimate is:

[0032] According to the survey results, all residents with electricity consumption less than or equal to 1 degree are not living. Therefore, the obtained Gaussian sub-distributions are sequentially sorted according to the mean of electricity consumption, and the first Gaussian sub-distribution with a mean greater than 1 degree is used as the electricity distribution of vacant houses, and the second Gaussian sub-distribution with a mean greater than 1 degree is used as the electricity distribution of houses with extremely low electricity consumption. The parameters of the Gaussian distribution of electricity consumption of vacant houses and houses with extremely low electricity consumption in the city are shown in Table 1: Table 1: Gaussian distribution parameters of vacant houses and houses with extremely low electricity consumption in the city

[0033] The data points with the same probability belonging to the two Gaussian sub-models, i.e., the intersection of the distribution of vacant state and the distribution of extremely low electricity consumption state, are calculated. And it is used as a segmentation point, and finally the threshold of monthly electricity consumption of vacant houses in the city is obtained. 8.37 degrees (as shown in Figure 2 , corresponding to 2.126, i.e., log(8.37+0.01)=2.126).

[0034] S102: Calculate the current incremental housing vacancy rate based on the electricity threshold of the housing vacancy state; ​In this step, the current incremental housing vacancy rate can be calculated by the following formula:

[0035] wherein, , represents the month when the incremental housing enters the market, for example represents January 2021; represents the month elapsed after the observation begins; represents the city In the number of incremental housing that flows into the market for sale in the month, represents the city In the first house of the incremental housing that enters the market in the month the monthly electricity consumption of the house from the month it flows into the market; represents the city From the proportion of vacant houses among sets of incremental housing that flow into the market for sale in the month.

[0036] In another exemplary embodiment, in step S200, the incremental housing vacancy rate prediction model is constructed, including the following steps: S201: Define a survival analysis framework; In this step, given that the incremental housing vacancy rate is between 0 and 1 and gradually decreases as the time to flow into the market increases, Figure 3 shows the incremental housing vacancy rate curve in a certain place in 2021 and 2022, which conforms to the characteristics of the survival curve, so the present application considers using the survival analysis curve to predict the change of the housing vacancy rate.

[0037] The present application introduces a survival analysis framework. In survival analysis, the survival function is between 0 and 1, and the survival probability is lower as time goes on. The curve based on the survival function can accurately depict the characteristics of the change of the incremental housing vacancy rate. The survival function of the exponential distribution fits the real characteristics of the gradual decrease of the incremental housing vacancy rate over time. Therefore, the present application selects the exponential distribution as the basis of the risk function model, which is specifically represented as follows:

[0038] wherein, represents the expected value of the actual observed vacancy rate under the condition ; represents parameters related to the entity and time , i.e., the city in Survival parameters of the month vary with city and observation start time; denotes the time step; is an exponential decay function, denoting a process that decays over time; denotes each city; denotes the month (observation start time) in which each city increment of housing enters the market, e.g. denotes January 2021; denotes the month elapsed since the observation start; denotes city in months after entering the market, the vacancy rate of the increment of housing in the actual observed vacancy rate the residual distribution of the model prediction value is normally distributed, and the present application assumes that all noises are independent. That is:

[0039] In practical applications, the survival parameters in the vacancy rate change curve of the increment of housing cannot be observed and vary with the observation start month, and therefore the present application describes the variation of the survival analysis parameters in the increment of housing vacancy rate model through a state space model.

[0040] S202: Under the framework of survival analysis, a state space model is constructed and the dynamic evolution process of the vacancy rate is described to associate the unobserved survival parameters with the observed data; In this step, the state space model includes state variables and measurement variables, and accordingly includes state equations and measurement equations. The measurement variables are observed variables, the state variables are variables that cannot be observed and change over time, and the state variables represent the running condition of a dynamic system at a certain time. The state equation describes the relationship between the state variables and the time, and the measurement equation describes the relationship between the measurement variables and the state variables. Thus, the state space model constantly simulates the true situation of the dynamic system through the measurement variables that can be observed. The dynamic system meets the conditions described by the state space model, and the observation equation and the state equation of the system are respectively:

[0041] wherein, denotes city in months after entering the market, the vacancy rate of the increment of housing in denotes the non-negative survival analysis parameter concerned by the present application, and denotes the noise subject to normal distribution with mean 0 and variance and Since has small variation over time, fix .

[0042] To ensure the non-negativity of the survival parameter, let Then the observation equation and state equation of the state space model are respectively:

[0043] where denotes the state variable of the entity at time , and denotes the vacancy rate of the housing at the current time; denotes the state variable of the entity at time .

[0044] The joint density function is:

[0045] where denotes the joint probability density function of the observation data and the state variable and the noise variance ; denotes the prior distribution of the state variable ; denotes the prior distribution of the noise variance ; denotes the product of the joint probability of all observation time points and each observation period in the time series; denotes the conditional probability of state transition, and denotes the conditional probability of the current state given the state at the last time.

[0046] S203: estimating the dynamic change of the survival parameter in the state space model based on the state space model and the Markov chain Monte Carlo algorithm, which specifically includes the following steps: S2031: determining the prior distribution of the parameter; In this step, it is assumed that the prior distribution of the survival parameter subject to normal distribution:

[0047] The prior distribution of the random error subject to inverse gamma distribution:

[0048] S2032: update and predict from the posterior distribution; update samples from the posterior distribution of state variables using Metropolis-Hastings algorithm update variance using Gibbs sampling algorithm predict the housing vacancy rate. The algorithm details are as follows, for each city :

[0049] To verify the rationality of the algorithm, according to the change rule of the vacancy rate curve, the present application considers the numerical simulation process as shown below:

[0050] wherein, represents the value at time , which is a random process; represents the value at time , which represents the lag of the random process; represents the random error term at time , which is usually set as a normal distribution; represents the initial value obeys a normal distribution with a mean of -3.5 and a variance of ; represents the error term obeys a normal distribution with a mean of 0 and a variance of 0.01; represents the output value at time and time step ; represents a nonlinear transformation based on an exponential decay function of ; represents an error term related to the output , which is usually assumed to be a normal distribution; represents the error term obeys a normal distribution with a mean of 0 and a variance of 0.01; , the initial vacancy rate of the simulated data is between 90.90% and 99.22%, and as increases, the overall vacancy rate shows a decreasing trend.

[0051] The MCMC algorithm is set as follows:

[0052] As a comparison, the most common autoregressive moving average model is considered in this application. The data is fitted and the change of the vacancy rate of houses is predicted on the same simulated data, whose formula is:

[0053] where, represents the predicted value at time ; represents the constant term; represents the regression coefficient related to the previous time point ; represents the order of the lag value in the model; represents the regression coefficient related to the previous error term ; represents the order of the error term lag in the model; represents the error term at the current time , and .

[0054] Table 2 is the vacancy rate of incremental housing of urban residents in a certain place from 2021 to 2024: Table 2

[0055] The data shown in Table 2 is taken as a data matrix, where the original data is regarded as an upper triangular matrix, and the blank part needs to be predicted. Through the MCMC algorithm, the estimated value of the survival parameter is obtained:

[0056] The survival parameter is brought into the survival function curve:

[0057] For a series of , the missing part in the matrix shown in Table 2 can be completed as the predicted value, as shown in Table 3: Table 3

[0058] Further, the mean square error is used to evaluate the prediction effect of the model, and the definition of MSE is:

[0059] where, represents the mean square error; represents the number of months elapsed since the observation from time , and ; represents the total number of samples; represents the first the time step of the sample, the actual observation value of the sample at time ; the predicted value of the sample at month ; The specific calculation process of the MSE is as follows: For the first column of observations , the fitted value is obtained by the ARMA model:

[0060] For the second column of observations , the predicted value is calculated by ARMR:

[0061] By analogy, the residual error of the sample can be calculated, and the MSE can be obtained by summing the squares of the residual error and taking the average.

[0062] Table 4 is the data of the vacancy rate of new electricity-connected houses from the time of electricity connection to 24 months after electricity connection in the first 45 months: Table 4

[0063] Table 5 is the prediction of the vacancy rate curve prediction model of the simulation data in each time period: Table 5

[0064] Based on the data shown in Tables 4 and 5, the MSE of the vacancy rate prediction model of the present application is 0.0002060594, and the MSE of the ARMA model is 0.003325503, which is 16 times the MSE of the state space model used in the present application, proving that the vacancy rate prediction model proposed in the present application has better prediction effect.

[0065] S204: Based on the model parameter estimation result, a survival analysis framework is combined to construct an incremental housing vacancy rate prediction model.

[0066] The incremental housing vacancy rate prediction model is represented as:

[0067] wherein, represents the predicted value of the variable corresponding to the entity at future time ; represents the predicted value of the variable corresponding to the entity and entity a rate of change under the entity; denotes the current time step; is an exponential function, representing a process that decays over time.

[0068] In another example embodiment, the present application also provides an incremental housing vacancy rate prediction device, comprising: an acquisition module 100 configured to acquire power data for use in housing vacancy rate prediction in a target area; a model construction module 200 configured to construct an incremental housing vacancy rate prediction model; and a prediction module 300 configured to use the power data as input of the incremental housing vacancy rate prediction model to predict an incremental housing vacancy rate in a future preset period.

[0069] In another example embodiment, the present application also provides a storage medium comprising instructions which, when executed on a computer, cause the computer to perform the method of any one of the preceding embodiments.

[0070] In another example embodiment, the present application also provides an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of the preceding embodiments when executing the program.

[0071] The above embodiments are only for illustrating the technical concepts and characteristics of the present application, and the purpose is to enable those skilled in the art to understand the content of the present application and implement it, and cannot limit the protection scope of the present application. Any equivalent changes or modifications made in accordance with the spirit and essence of the present application shall be covered within the protection scope of the present application.

Claims

1. A method for predicting incremental housing vacancy rates, characterized in that, The prediction method comprises: acquiring power data in a target area for housing vacancy rate prediction; constructing an incremental housing vacancy rate prediction model; using the power data as input of the incremental housing vacancy rate prediction model to predict an incremental housing vacancy rate in a future preset period.

2. The method of claim 1, wherein, The acquiring of the power data in the target area for the housing vacancy rate prediction comprises: acquiring monthly electricity consumption data of incremental housing of urban residents in the target area, fitting the monthly electricity consumption data according to a Gaussian mixture model to obtain a housing vacancy state electricity consumption threshold; calculating a current incremental housing vacancy rate based on the housing vacancy state electricity consumption threshold.

3. The method of claim 1, wherein, The constructing of the incremental housing vacancy rate prediction model comprises: defining a survival analysis framework; under the survival analysis framework, constructing a state space model and describing a dynamic evolution process of the vacancy rate to associate unobserved survival parameters with observed data; based on the state space model, estimating dynamic changes of the survival parameters in the state space model through a Markov chain Monte Carlo algorithm; based on a model parameter estimation result, combining the survival analysis framework to construct an incremental housing vacancy rate prediction model.

4. The method of claim 3, wherein, The estimation of the dynamic changes of the survival parameters in the state space model through the Markov chain Monte Carlo algorithm based on the state space model comprises: determining a parameter prior distribution; sampling and updating state variables from a posterior distribution of the state variables through a Metropolis-Hastings algorithm, updating a variance through Gibbs sampling to obtain an estimation result of parameters of the state space model.

5. The method of claim 3, wherein, The incremental housing vacancy rate prediction model is expressed as: in, Indicates at time At that time, entity Corresponding variables In the future The predicted value; Indicates at time and entity Rate of change below; Indicates the current time step; It is an exponential function, representing a process that decays over time.

6. A time series analysis-based incremental housing vacancy rate prediction device characterized by comprising: The device comprises: an acquisition module configured to acquire power data in a target area for housing vacancy rate prediction; a model construction module configured to construct an incremental housing vacancy rate prediction model; a prediction module configured to use the power data as input of the incremental housing vacancy rate prediction model to predict an incremental housing vacancy rate in a future preset period.

7. A storage medium, characterized by The storage medium comprises instructions which, when executed on a computer, cause the computer to perform the method according to any one of claims 1 to 5.

8. An electronic device, comprising: The electronic device comprises a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 5 when executing the program.

Images