Urban update potential evaluation method based on building information and human activity intensity

Abstract

The invention discloses a city update potential evaluation method based on building information and human activity intensity, and relates to the field of city update potential evaluation, and the method comprises the steps: constructing a city update potential evaluation index system taking building information and human activity intensity information into consideration, and carrying out the quantitative evaluation of all indexes based on the block-level quantitative scale; based on a quantitative evaluation result, constructing a city updating single classification data set, and training a classifier through the city updating single classification data set to obtain a city updating probability model; correcting the probability model through the correction normal number to obtain a city update potential evaluation model, and outputting a city update evaluation result of each block by using the city update potential evaluation model; and explaining the output of the city update potential evaluation model by using a model explaining tool to obtain a feature contribution value, and evaluating a driving relationship between city update and each input feature based on the feature contribution value. According to the invention, city update potential evaluation of the block scale is realized.

Description

Technical Field

[0001] The invention relates to the field of urban renewal potential evaluation, and in particular to an urban renewal potential evaluation method based on building information and human activity intensity. Background Art

[0002] Urbanization has created tremendous value for economic and social development and significantly improved the quality of life of residents. There are various types of urban renewal. According to the strength of the transformation, it can be divided into three types: remediation, reconstruction and demolition and reconstruction. According to the different objects of remediation, it can be divided into residential renewal, commercial renewal, industrial renewal and other renewal. The evaluation of urban renewal potential is to determine the priority and key areas of urban renewal practice through a comprehensive evaluation of social economy, etc. The evaluation of urban renewal potential is a key link in conducting urban renewal research. It can provide a scientific basis for urban renewal decision-making and help cities achieve high-quality and sustainable development.

[0003] The evaluation of urban renewal potential often requires the integration of multi-source data. The data sources used in existing research usually include survey data, statistical data, GIS (Geographic Information System) spatial data, etc. Survey data are usually obtained through field surveys, interest group survey interviews, expert interviews, etc. Data survey work is usually carried out passively and discretely with the renewal project as the guide. The collection cost is high and the efficiency is low, which is not conducive to a comprehensive grasp of the overall situation of the urban renewal scope. It is currently a common practice to obtain relevant factors by combining statistical data and GIS spatial data, which can make up for the shortcomings of field survey data. However, statistical data and GIS spatial data usually record the plane attribute information of the land at a certain moment or a certain time interval. The rapid development of geographic big data such as mobile phone signaling, Weibo check-in, and taxi trajectories can provide information on the spatial distribution differences of human activity intensity, and provide data support for capturing the internal characteristics of the city from a fine scale and dynamic perspective. At the same time, the emergence of three-dimensional building model data provides the possibility of understanding urban characteristics from a vertical dimension.

[0004] The evaluation method based on the multi-factor and multi-factor comprehensive discrimination model is currently the most commonly used decision support method that integrates multi-source information. The multi-factor and multi-factor discrimination model usually selects relevant indicators to evaluate the suitability of the region and then determines the decision plan. The indicator system usually varies according to the research perspective, evaluation object, etc. For example, the research on urban sustainable development mainly selects indicators from the aspects of social, economic, environmental, and architectural conditions. However, there is a disadvantage of strong subjectivity when making urban renewal decisions based on the multi-factor and multi-factor comprehensive discrimination model. First, when the indicators are integrated, it is necessary to set the indicator weights. Commonly used indicator weighting methods include hierarchical analysis method, entropy method or fuzzy logic method, all of which show strong subjectivity. Second, the level division of potential evaluation results is usually subjective, for example: the potential results are divided into two levels and different models of urban renewal areas are determined accordingly. These two types of subjective problems will greatly affect the potential evaluation results of urban renewal.

[0005] In order to overcome the subjective shortcomings of the multi-factor and multi-factor comprehensive discrimination method, a machine learning method can be introduced to obtain the relationship between urban renewal potential and related indicators, which has high efficiency and objectivity. Under normal circumstances, the use of binary classifiers or multi-classifiers can identify urban areas of interest. However, areas where urban renewal is not carried out have both land with high suitability for urban renewal and land with low suitability for urban renewal, that is, they contain both positive samples and negative samples. It is difficult and subjective to mark land (negative samples) where urban renewal is not carried out at all. Therefore, the use of binary classifiers or multi-classifiers to identify urban renewal potential has limitations.

[0006] Single classifiers can solve the problem of difficult negative sample labeling. In remote sensing classification, single classifiers are used to detect a certain class of objects of interest and reject objects of other classes. However, the model is very sensitive to the choice of parameters, and users need to adjust the parameter thresholds through experience to balance over-prediction and under-prediction. In addition, in the absence of negative samples, these single classifiers need enough positive samples to estimate the boundaries of the target class. Summary of the invention

[0007] In view of the problems in the related technologies, the present invention proposes an urban renewal potential evaluation method based on building information and human activity intensity to overcome the above-mentioned technical problems existing in the existing related technologies.

[0008] To this end, the specific technical solution adopted by the present invention is as follows:

[0009] The urban renewal potential evaluation method based on building information and human activity intensity includes:

[0010] S1. Construct an urban renewal potential evaluation index system that takes into account building information and human activity intensity information, and conduct quantitative evaluation of each index based on a block-level quantitative scale.

[0011] S2. Based on the results of quantitative evaluation, a single-classification dataset of urban renewal is constructed, and a classifier is trained using the single-classification dataset to obtain a probabilistic model of urban renewal.

[0012] S3. Correct the probability model by using the correction constant to obtain the urban renewal potential evaluation model, and use the urban renewal potential evaluation model to output the urban renewal evaluation results of each block.

[0013] S4. Use the model interpretation tool to interpret the output of the urban renewal potential evaluation model to obtain the characteristic contribution value, and evaluate the driving relationship between urban renewal and each input feature based on the characteristic contribution value.

[0014] Furthermore, the urban renewal potential evaluation index system includes the dimensions of socio-economic conditions, building conditions, land use conditions and natural environment conditions.

[0015] Furthermore, based on the results of quantitative evaluation, a single-classification dataset of urban renewal is constructed, and the classifier is trained through the single-classification dataset of urban renewal. The probability model of urban renewal is obtained, including:

[0016] S21. Determine input features according to the results of quantitative evaluation, and use the input features and labels as samples to form a single-classification data set;

[0017] S22. By randomly extracting samples from the single-classification data set, a training sample set is obtained, and a binary classifier is trained using the training sample set to obtain a probability model of urban renewal.

[0018] Furthermore, the samples include positive samples and unlabeled samples;

[0019] The positive samples are of known categories; the unlabeled samples are of unknown categories.

[0020] Furthermore, the probability model is corrected by the correction constant to obtain the urban renewal potential evaluation model, and the urban renewal potential evaluation model is used to output the urban renewal evaluation results of each block, including:

[0021] S31, obtaining a typical existing sample set, and substituting all the typical existing samples into the probability model to obtain an output value; and obtaining a correction constant by calculating the average of all the output values;

[0022] S32, calibrating the probability model by using a calibration constant, and using the calibrated probability model as an urban renewal potential evaluation model;

[0023] S33. Input the predicted data set into the urban renewal potential evaluation model to obtain the probability of occurrence of urban renewal potential.

[0024] Furthermore, obtaining a typical existing sample set includes:

[0025] By randomly extracting several positive samples from the training sample set, a typical existence sample set is obtained.

[0026] Furthermore, the urban renewal potential evaluation model is:

[0027]

[0028] Where c represents the calibration constant;

[0029] x i represents the i-th input feature;

[0030] y represents the sample;

[0031] P(s=1|x i ,η=1) represents the probability model;

[0032] s=1 represents a positive sample, and η=1 represents a sampling method in which positive samples and unlabeled samples are collected separately.

[0033] Furthermore, the probability model is corrected by the correction constant to obtain the urban renewal potential evaluation model, and the urban renewal evaluation results of each block are output by using the urban renewal potential evaluation model, including:

[0034] The accuracy of the urban renewal potential evaluation model was calculated using true positive, false negative and false positive indicators;

[0035] The root mean square error was selected as the accuracy verification indicator of the urban renewal potential evaluation model.

[0036] Furthermore, the calculation formula for calculating the accuracy of the urban renewal potential evaluation model using true positive, false negative and false positive indicators is:

[0037]

[0038] Wherein, TP', FN' and FP' represent the true positive index, false negative index and false positive index respectively.

[0039] Furthermore, the output of the urban renewal potential evaluation model is interpreted using a model interpretation tool to obtain the characteristic contribution value, and the driving relationship between urban renewal and various input characteristics is evaluated based on the characteristic contribution value, including:

[0040] S41. Use SHAP technology to calculate the characteristic contribution value of each input feature in the urban renewal potential evaluation model;

[0041] S42. Use the feature contribution value to evaluate the influence of each input feature on the output of the urban renewal potential evaluation model, and obtain the driving relationship between urban renewal and each input feature based on the influence of each input feature.

[0042] Beneficial effects of the present invention:

[0043] 1. This paper integrates multi-source geographic big data, building three-dimensional data, basic geographic data, statistical data, etc., takes the block as the evaluation unit, and constructs 28 indicators in four aspects including natural environment conditions, land use conditions, socio-economic conditions, and building conditions from the vertical dimension, dynamic angle, and dynamic angle. Taking the existing urban renewal practice area as the positive sample, four types of urban renewal single classification data sets, including residential, industrial, commercial, and other types, are generated, and PBL is constructed respectively. RF The urban renewal potential evaluation model realizes the urban renewal potential evaluation at the block scale. Urban renewal potential evaluation is an important means to make sustainable urban renewal decisions.

[0044] 2. Through PBL RF Combined with SHAP analysis, the complex driving relationships between various factors and urban renewal, such as primary and secondary, positive and negative, and superposition, can be obtained. The present invention can achieve objective and accurate evaluation of urban renewal potential and provide scientific support for research on sustainable urban renewal decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. 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 paying creative work.

[0046] Figure 1 A flow chart of a method for evaluating urban renewal potential based on building information and human activity intensity according to an embodiment of the present invention;

[0047] Figure 2 This is a flow chart of urban renewal potential evaluation based on PBL according to an embodiment of the present invention;

[0048] Figure 3 A schematic diagram of driving factors for residential urban renewal according to an embodiment of the present invention;

[0049] Figure 4 This is a schematic diagram of driving factors for industrial urban renewal according to an embodiment of the present invention;

[0050] Figure 5 This is a schematic diagram of driving factors for commercial urban renewal according to an embodiment of the present invention;

[0051] Figure 6 Schematic diagram of other types of urban renewal driving factors according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. However, it should be understood that the present invention can be implemented in various forms, and some exemplary and non-limiting embodiments shown in the accompanying drawings and described below are not intended to limit the present invention to the specific embodiments described.

[0053] It should be understood that, where technically feasible, the technical features listed above for different embodiments can be combined with each other to form other embodiments within the scope of the present invention. In addition, the specific examples and embodiments described in the present invention are non-limiting, and the structures, steps, and sequences described above can be modified accordingly without departing from the scope of protection of the present invention.

[0054] According to an embodiment of the present invention, a method for evaluating the potential of urban renewal based on building information and human activity intensity is provided. Limited by traditional data sources, the existing technology rarely comprehensively considers multidimensional urban characteristics such as urban building three-dimensional information and human activity intensity when evaluating the potential of urban renewal. Therefore, it is necessary to integrate three-dimensional building data and multi-source geographic big data, etc., and construct 28 indicators in four aspects including natural environmental conditions, land use, socio-economic conditions, and building conditions from a vertical dimension, dynamic perspective, and fine scale; and use the existing urban renewal practice area as a positive sample to construct an urban renewal potential evaluation model based on the presence and background learning and random forest (PBL-RF).

[0055] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1-2 As shown, according to the urban renewal potential evaluation method based on building information and human activity intensity according to an embodiment of the present invention, the urban renewal potential evaluation method based on building information and human activity intensity includes:

[0056] S1. Construct an urban renewal potential evaluation index system that takes into account building information and human activity intensity information, and conduct quantitative evaluation of each index based on a block-level quantitative scale.

[0057] In one embodiment, referring to the research on urban sustainable development and urban renewal decision support, 28 indicators are selected from four dimensions, namely, socio-economic conditions, building conditions, land use conditions, and natural environmental conditions, to construct an urban renewal potential evaluation index system (Table 1). The quantitative scale of each indicator is the block scale. The specific connotation, quantitative method, and quantitative results of each single indicator are as follows:

[0058] Table 1 Urban renewal potential evaluation index system

[0059]

[0060]

[0061] (1) Socioeconomic conditions

[0062] Population density can measure the state of land abandonment or overpopulation. Five indicators, namely the distance to parks / open spaces, the number of educational and medical facilities, the number of enterprises, the number of living service facilities, and the number of transportation facilities, affect the value of block land, the convenience of residents' lives and employment, etc., and have a great impact on the potential for urban renewal. The rental level of urban villages, old factories, etc. will be relatively low, which is closely related to urban renewal. The average rent of different land types is calculated, and the larger the value, the higher the rent level. The number of patent applications represents the ability of industry to create efficiently. Old factories usually have fewer patent applications and are more likely to be renewed in the city.

[0063] Taxi GPS data is needed to quantify the four indicators of work-life balance, average travel distance during peak travel periods, degree of neighborhood connectivity, and nighttime economic activities. Urban areas with imbalanced work-life balance face problems such as long commuting distances and long commuting times. Urban renewal will help promote the integration of industry and cities. The proportion of taxi orders in each block with a travel time of less than 30 minutes during peak commuting periods is used to quantify "work-life balance". Among them, the peak commuting periods occur between 8:00-10:00 and 17:00-19:00.

[0064] Existing studies have shown that due to the lack of sufficient professional skills, urban village residents are mostly employed in nearby labor-intensive, low-tech industrial land, and their commuting distances are usually short. The average taxi outbound distances during the commuting peak period of different land types were counted as the quantitative value of "average outbound distance during the travel peak period", and the larger the value, the longer the outbound distance. Among them, the peak commuting hours for different land types are different, with residential land being 17:00-19:00, industrial land being 8:00-10:00, commercial land being 18:00-20:00, and public service land being 14:00-16:00.

[0065] The degree of block connection and nighttime economic activity can characterize the vitality of the block from a dynamic perspective, and then characterize the abandoned state of the land. The present invention uses the frequency of taxis entering the block as a standard for measuring the degree of block connection, and the larger the value, the more frequent the entry. The taxi trip orders at night (20:00 to 6:00 the next day) are used as a standard for measuring nighttime economic activities, and the larger the value, the more nighttime economic activities.

[0066] The present invention defines temporal diversity as the temporal distribution richness of population activities. Weibo check-in data has high temporal accuracy, and the present invention uses its distribution characteristics in different time periods to calculate temporal diversity. The Weibo check-in data is divided into five time periods: morning (6:00-10:00), noon (10:00-14:00), afternoon (14:00-17:00), evening (17:00-22:00), and late night (22:00-06:00). Using the formula Calculate the time diversity, where T i Represents the proportion of Weibo check-ins in different time periods. SHDI time A larger quantization value represents a greater temporal diversity.

[0067] (2) Building conditions

[0068] Based on the building data from a city's Land and Resources Commission, the building situation at the block scale can be quantified from a fine scale and vertical dimension. The average building age and average building height can be quantified by the construction date and building floor attributes. Building density is the ratio of the building base area in the block to the block area. The floor area ratio is the ratio of the total building area in the block to the block area. The three-dimensional urbanization index can reflect the progress of urbanization and quantify it by building height and building base area (3DUI = ∑ n A i ×H i , 3DUI is the three-dimensional urbanization index of the block, A i is the building base area, H i is the building height, and n is the number of buildings in the block). Per capita building area is an important basis for urban planning and urban layout, which is the ratio of the building area of ​​the block to the population.

[0069] (3) Land use

[0070] Land mix has an important impact on urban sustainable development. Diversity can reflect the quantitative characteristics of land in a block, accessibility can reflect the spatial distribution of land, and compatibility can reflect the interaction between lands. It is an effective indicator to measure the potential of urban renewal.

[0071] Based on the land survey data, the two indicators of "property owners involved" and "land ownership" are quantified. According to the requirements of the urban renewal policy, the implementation of urban renewal projects must obtain the consent of all relevant land and building owners. A project involving fewer property owners has lower transaction costs in the reconstruction process and is more likely to be redeveloped. Using land survey information and block boundaries, information on property owners involved in each block can be obtained. Land ownership is mainly divided into three categories: state-owned land, collective land and mixed land. Projects with state-owned land have lower transaction costs and are more likely to be redeveloped than projects with collective land and mixed land. The proportion of state-owned land in each block is counted as a quantitative indicator of land ownership.

[0072] Urban land types are divided according to natural landscape, land use function, human activities, and building conditions. The quantitative scores of dense low-rise industrial land and open low-rise industrial land are 10, and the quantitative scores of open low-rise residential land, dense low-rise residential land, and low shared service land are 8, 6, and 4 respectively. The quantitative scores of open low-rise commercial land and dense low-rise commercial land are 2. The quantitative scores of other urban land types are 0.

[0073] (4) Natural environmental conditions

[0074] Good environmental conditions have a positive impact on residents' work, life, and residence. The greening rate is the ratio of green area to block area, which can reflect the environmental conditions of the block. Elevation affects the land use layout of buildings and the difficulty of engineering, and is one of the important factors affecting the choice of urban land use. Enterprise pollution information has an important impact on industrial renewal. The more polluting the block is, the greater the possibility of redevelopment. In the evaluation results of a certain city, there are 171 green-card enterprises with the least pollution, 757 blue-card enterprises with less pollution, 99 yellow-card enterprises with more pollution, and 57 red-card enterprises with extremely high pollution. The green card, blue card, yellow card, and red card are assigned values ​​of 5, 4, 3, and 1 respectively, and the evaluation results of 1,084 enterprises are interpolated by inverse distance weights to calculate the average pollution level of each block.

[0075] To establish an urban renewal potential evaluation model based on a single classifier, it is necessary to construct a single classification data set, including positive samples (s=1) and unlabeled samples (s=0). The input features are obtained according to the quantitative indicators in S1. Secondly, an urban renewal potential evaluation model is constructed based on the single classification data set and PBL. Model construction includes two parts: model training and accuracy verification. Finally, the driving factors of urban renewal and the potential evaluation results are analyzed according to the potential evaluation model (such as Figure 2 shown).

[0076] S2. Based on the results of quantitative evaluation, a single-classification dataset of urban renewal is constructed, and a classifier is trained using the single-classification dataset to obtain a probabilistic model of urban renewal.

[0077] Let x be the geographic covariate (input feature) associated with the sample label. For common binary classification data {x i , y} includes positive samples (y = 1) and negative samples (y = -1). In common cases, the data that can be used for training is usually an incomplete set of binary classification samples, that is, only some positive samples are labeled and negative samples are completely unlabeled. Such a data set is called single-classification data {x i , s}. Among them, s∈{1,0} indicates whether the sample is assigned a label. If the class of x is clearly known, it is called a labeled positive sample (s=1). If the class of x is unknown, it is called an unlabeled sample (s=0). That is, when s=1, y=1; when s=0, y=1 or y=-1. The present invention refers to the positive sample-unlabeled sample dataset as Geo-one class data (GOCD).

[0078] For example, vectorize the urban renewal demolition and construction range to obtain block-scale urban renewal data. Mark the blocks that have been urban renewed as positive samples (s=1). Input feature x i According to the quantitative indicators in S1, a total of 28 characteristic indicators are obtained, including: socio-economic conditions, building conditions, land use conditions, and natural environment conditions. Unlabeled samples can be randomly selected from the remaining blocks.

[0079] In one embodiment, based on the results of the quantitative evaluation, a single-classification dataset of urban renewal is constructed, and a classifier is trained using the single-classification dataset of urban renewal to obtain a probability model of urban renewal, including:

[0080] S21. Determine input features according to the results of quantitative evaluation, and use the input features and labels as samples to form a single-classification data set;

[0081] S22, randomly extract samples from the single-classification data set to obtain a training sample set, and use the training sample set to train a binary classifier to obtain a probability model P (s = 1 | x i , η=1). Among them, s=1 represents the labeled positive sample, that is, the positive sample of the urban renewal block; s=0 represents the unlabeled sample; η=1 represents the sampling method in which the positive sample and the unlabeled sample are collected separately, that is, the positive sample is first randomly selected from the sample set, and then the unlabeled sample is randomly selected separately. Since the unlabeled data (s=0) contains positive samples (y=1) and negative samples (y=-1), it is not the real negative sample data, so P(s=1|x i , η=1) can distinguish s=0 and s=1, but cannot distinguish y=1 and y=-1, that is: P(s=1|x i , η=1) and the required P(y=1|x i) are different.

[0082] In one embodiment, the samples include positive samples and unlabeled samples;

[0083] The positive samples are of known categories; the unlabeled samples are of unknown categories.

[0084] S3. Correct the probability model by using the correction constant to obtain the urban renewal potential evaluation model, and use the urban renewal potential evaluation model to output the urban renewal evaluation results of each block.

[0085] In one embodiment, the probability model is corrected by a correction constant to obtain an urban renewal potential evaluation model, and the urban renewal potential evaluation model is used to output the urban renewal evaluation results of each block, including:

[0086] S31. Obtain a typical existing sample set, and substitute all typical existing samples into the probability model to obtain output values; obtain the correction constant by calculating the average of all output values:

[0087]

[0088] Where O is a subset of labeled positive samples (s = 1), which is a typical existence point. “Typical existence” means that the block is 100% suitable for urban renewal. c is the P(s = 1|x i ,η=1). n is the number of data sets O.

[0089] S32. Correct the probability model by using the correction constant, and use the corrected probability model as the urban renewal potential evaluation model.

[0090] S33. Input the predicted data set into the urban renewal potential evaluation model to obtain the probability of occurrence of urban renewal potential.

[0091] To construct PBL, you need to select a binary classifier (or multiple classifiers) that can estimate conditional probabilities. The present invention uses the random forest classifier (RF) mainly because it not only has high classification accuracy, but also can be well coupled with the SHAP model to identify the key factors that affect the model. RF After that, the output function value can be normalized to fall between 0 and 1. The present invention randomly extracts 50% of the data from GOCD-UR (Geo-one class data of Urban Renewal) for PBL RF Model training (GOCD-UR1) (such as Figure 2As shown in the figure, the remaining data (GOCD-UR2) is used as a validation data set and does not participate in the training of the model. In order to improve the generalization ability of the model and avoid the error of c value estimation, 15% of positive samples are randomly selected from GOCD-UR1 as typical existence samples (O data set). Finally, the prediction data set is input into the model PBL RF The urban renewal potential evaluation result P(y=1|x i ), the value range is [0, 1].

[0092] In one embodiment, obtaining a typical presence sample set includes:

[0093] By randomly extracting several positive samples from the training sample set, a typical existence sample set is obtained.

[0094] In one embodiment, the urban renewal potential evaluation model is:

[0095]

[0096] Where c represents the calibration constant;

[0097] x i represents the i-th input feature;

[0098] y represents the sample;

[0099] P(s=1|x i ,η=1) represents the probability model; s=1 represents the positive sample, and η=1 represents the sampling method in which the positive sample and the unlabeled sample are collected separately.

[0100] Traditional evaluation indicators (such as Kappa coefficient, F-measure, mapping accuracy, user accuracy, etc.) need to calculate TP (i.e., the number of positive samples correctly identified), TN (the number of negative samples correctly identified), FP (the number of negative samples misclassified as positive samples) and FN (the number of positive samples that are missed). However, the GOCD-UR of the present invention is single-classification data and lacks negative samples, so it cannot be directly used for the calculation of traditional indicators.

[0101] In one embodiment, the probability model is corrected by a correction constant to obtain an urban renewal potential evaluation model, and the urban renewal potential evaluation model is used to output the urban renewal evaluation results of each block, which includes:

[0102] The accuracy of the urban renewal potential evaluation model is calculated using true positive, false negative and false positive indicators, which can correctly evaluate the accuracy of the single classifier model in the absence of negative samples;

[0103] The root mean square error (RMSE) was selected as the accuracy verification indicator of the urban renewal potential evaluation model.

[0104]

[0105] Where Q is all the labeled positive samples in the validation dataset, and m represents the number of samples in the Q dataset. Q (y=1|x i ) represents the label in the Q dataset, and its value is 1. P(y=1|x i ) is the model PBL RF The prediction result is in the interval [0, 1]. The smaller the RMSE value, the higher the model accuracy.

[0106] In one embodiment, the calculation formula for calculating the accuracy of the urban renewal potential evaluation model using true positive, false negative and false positive indicators is:

[0107]

[0108] In the formula, TP', FN' and FP' represent the true positive index, false negative index and false positive index respectively. pb It is a special verification method for PBL model, which can be calculated by confusion matrix. The larger the value, the higher the model accuracy. To construct the confusion matrix, P(y=1|x i ) is binarized, and the threshold is selected as 0.4. If P(y=1|x i )≥0.4, then the predicted value y=1;

[0109] P(y=1|x i )<0.4, the predicted value y=-1. In addition, in the construction of the confusion matrix, the unlabeled samples are regarded as "pseudo" negative samples.

[0110] Model training was performed for industrial urban renewal, residential urban renewal, commercial urban renewal, and other types of renewal (Table 2). There were 1925 training samples (333 labeled positive samples and 1592 unlabeled samples), with the largest number of training samples for industrial renewal (253 labeled positive samples and 1247 unlabeled samples); there were 1925 validation samples (329 labeled positive samples and 1596 unlabeled samples). The accuracy of the results showed that the evaluation results of industrial urban renewal potential were the most accurate, followed by residential and commercial types, and the evaluation results of other types of urban renewal potential were the least accurate.

[0111] Table 2 Model training accuracy

[0112]

[0113] The four urban renewal potential evaluation models constructed have a potential evaluation result score P(y=1|x i) all fall between 0 and 1. The equal interval method is used to divide the urban renewal potential evaluation results into five levels: extremely low potential blocks [0, 0.2], low potential blocks [0.2, 0.4], medium potential blocks [0.4, 0.6], high potential blocks [0.6, 0.8], and extremely high potential blocks [0.8, 1.0]. The number of blocks in different urban renewal potential levels is counted, and it is found that the proportions of blocks in potential levels [0, 0.2], [0.2, 0.4], [0.4, 0.6], [0.6, 0.8], and [0.8, 1.0] are 56%, 22%, 9%, 5%, and 8%, respectively.

[0114] RF is a common machine learning algorithm that performs well in both classification and regression problems. However, RF is often considered a black box algorithm and its prediction results are difficult to explain. Therefore, in order to better explain the prediction results of RF, the SHAP (SHapley Additive exPlanations) technique is used. SHAP is a method for explaining model predictions. It is based on the concept of Shapley values ​​and provides a comprehensive method to explain the output of any machine learning model.

[0115] S4. Use the model interpretation tool (SHAP technology) to interpret the output of the urban renewal potential evaluation model to obtain the characteristic contribution value, and evaluate the driving relationship between urban renewal and each input feature based on the characteristic contribution value.

[0116] In one embodiment, the output of the urban renewal potential evaluation model is interpreted using a model interpretation tool to obtain a feature contribution value, and the driving relationship between urban renewal and each input feature is evaluated based on the feature contribution value, including:

[0117] S41. Using SHAP technology to calculate the characteristic contribution value (SHAP value) of each input feature in the urban renewal potential evaluation model can help us understand how each feature affects the prediction results of the RF model, thereby better understanding the prediction results of the model;

[0118] S42. Use the feature contribution value to evaluate the influence of each input feature on the output of the urban renewal potential evaluation model, and obtain the driving relationship between urban renewal and each input feature based on the influence of each input feature.

[0119] Using SHAP to calculate P(s=1|x i,η=1) model to explain and then analyze the driving factors of different types of urban renewal. First, the main driving factors of different types of urban renewal are different. Land type, average building age, population density and average driving distance during peak travel have a significant positive impact on residential urban renewal; on the contrary, average building height, floor area ratio, land ownership, three-dimensional urbanization index, rent and accessibility have a significant negative impact on residential urban renewal. The greater the compatibility, three-dimensional urbanization index, average building height, number of patent applications, per capita building area, polluting enterprises, floor area ratio, population density and number of owners involved, the less likely it is to carry out industrial urban renewal; on the contrary, the greater the accessibility, greening rate, proportion of state-owned land, average building age, elevation, building density and land type quantitative value, the greater the potential for industrial urban renewal. The greater the time diversity, land mix diversity, block connection, average building height, three-dimensional urbanization index and floor area ratio, the less likely it is to carry out commercial urban renewal; on the contrary, the greater the land type, population density and accessibility, the greater the possibility of carrying out commercial urban renewal. Indicators such as population density, three-dimensional urbanization index, average building height, average building age and number of owners involved have a significant negative impact on other types of urban renewal, while land type has a significant positive impact on other types of urban renewal.

[0120] Secondly, the same driving factors have different positive and negative effects on different urban renewal. In land ownership, the larger the proportion of state-owned land area, the smaller the potential for residential urban renewal, and the greater the potential for industrial urban renewal. This is mainly because residential urban renewal mostly occurs in urban villages, which have more collective land and mixed land, and the proportion of state-owned land area is small. On the contrary, the transaction cost of industrial urban renewal in areas with a small proportion of state-owned land is low, which is conducive to the development of industrial urban renewal. The greater the accessibility, the greater the potential for industrial and commercial urban renewal, and the smaller the potential for residential urban renewal. Carrying out industrial and commercial urban renewal in areas with good accessibility can bring more benefits, while residential areas with good accessibility have good social and economic conditions and usually do not need further optimization. In addition, population density has different positive and negative effects on urban renewal of different land use types. Industrial urban renewal mainly occurs in areas with low population density, and residential and commercial urban renewal mainly occurs in areas with high population density.

[0121] There are complex driving relationships between various factors and different types of urban renewal, such as primary and secondary, positive and negative, and superposition. The geographic big data and three-dimensional building data that represent the intensity of human activities have a significant impact on the potential evaluation results.

[0122] Schematic diagram of the driving factors for residential urban renewal, industrial urban renewal, commercial urban renewal, and other urban renewal factors, such as Figure 3-Figure 6 As shown in the figure, Mean(SHAP value) represents the mean of the SHAP value, and average impact on model output magnitude represents the average magnitude of the impact on the model output. SHAP value represents the SHAP value, and impact on model output represents the impact on the model output.

[0123] In summary, this paper integrates multi-source geographic big data, building three-dimensional data, basic geographic data, statistical data, etc., takes the block as the evaluation unit, and constructs 28 indicators in four aspects including natural environment conditions, land use conditions, socio-economic conditions, and building conditions from the vertical dimension, dynamic angle, and dynamic angle. Taking the existing urban renewal practice area as the positive sample, four types of urban renewal single classification data sets, including residential, industrial, commercial, and other types, are generated, and PBL is constructed respectively. RF The urban renewal potential evaluation model realizes the urban renewal potential evaluation at the block scale. Urban renewal potential evaluation is an important means to make sustainable urban renewal decisions. RF Combined with SHAP analysis, the complex driving relationships between various factors and urban renewal, such as primary and secondary, positive and negative, and superposition, can be obtained. The present invention can achieve objective and accurate evaluation of urban renewal potential and provide scientific support for research on sustainable urban renewal decision-making.

[0124] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0125] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. An urban renewal potential evaluation method based on building information and human activity intensity, characterized in that: The urban renewal potential evaluation method based on building information and human activity intensity includes: S1. Construct an urban renewal potential evaluation index system that takes into account building information and human activity intensity information, and conduct quantitative evaluation of each index based on the block-level quantitative scale; S2. Based on the results of quantitative evaluation, a single-classification dataset of urban renewal is constructed, and a classifier is trained using the single-classification dataset of urban renewal to obtain a probability model of urban renewal. S3, correcting the probability model by using the correction constant to obtain the urban renewal potential evaluation model, and using the urban renewal potential evaluation model to output the urban renewal evaluation results of each block; S4. Use the model interpretation tool to interpret the output of the urban renewal potential evaluation model to obtain the characteristic contribution value, and evaluate the driving relationship between urban renewal and each input feature based on the characteristic contribution value.

2. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 1 is characterized in that: The urban renewal potential evaluation index system includes the dimensions of socio-economic conditions, building conditions, land use conditions and natural environment conditions.

3. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 1 is characterized in that: Based on the results of the quantitative evaluation, a single-classification data set for urban renewal is constructed, and a classifier is trained using the single-classification data set for urban renewal to obtain a probability model for urban renewal, including: S21. Determine input features according to the results of quantitative evaluation, and use the input features and labels as samples to form a single-classification data set; S22. By randomly extracting samples from the single-classification data set, a training sample set is obtained, and a binary classifier is trained using the training sample set to obtain a probability model of urban renewal.

4. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 3 is characterized in that: The samples include positive samples and unlabeled samples; The positive samples are of known categories; the unlabeled samples are of unknown categories.

5. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 1 is characterized in that: The method of correcting the probability model by using the correction constant to obtain the urban renewal potential evaluation model and outputting the urban renewal evaluation results of each block by using the urban renewal potential evaluation model includes: S31, obtaining a typical existing sample set, and substituting all the typical existing samples into the probability model to obtain an output value; and obtaining a correction constant by calculating the average of all the output values; S32, calibrating the probability model by using a calibration constant, and using the calibrated probability model as an urban renewal potential evaluation model; S33. Input the predicted data set into the urban renewal potential evaluation model to obtain the probability of occurrence of urban renewal potential.

6. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 5 is characterized in that: The obtaining of a typical existing sample set comprises: By randomly extracting several positive samples from the training sample set, a typical existence sample set is obtained.

7. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 5 is characterized in that: The urban renewal potential evaluation model is: Where c represents the calibration constant; x i represents the i-th input feature; y represents the sample; P(s=1|x i ,η=1) represents the probability model; s=1 represents a positive sample, and η=1 represents a sampling method in which positive samples and unlabeled samples are collected separately.

8. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 5 is characterized in that: The method comprises the following steps: correcting the probability model by using the correction constant to obtain the urban renewal potential evaluation model, and outputting the urban renewal evaluation results of each block by using the urban renewal potential evaluation model: The accuracy of the urban renewal potential evaluation model was calculated using true positive, false negative and false positive indicators; The root mean square error was selected as the accuracy verification indicator of the urban renewal potential evaluation model.

9. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 8 is characterized in that: The calculation formula for calculating the accuracy of the urban renewal potential evaluation model using true positive, false negative and false positive indicators is: Wherein, TP', FN' and FP' represent the true positive index, false negative index and false positive index respectively.

10. The urban renewal potential evaluation method based on building information and human activity intensity according to claim 1 is characterized in that: The output of the urban renewal potential evaluation model is interpreted using a model interpretation tool to obtain a characteristic contribution value, and the driving relationship between urban renewal and various input characteristics is evaluated based on the characteristic contribution value, including: S41. Use SHAP technology to calculate the characteristic contribution value of each input feature in the urban renewal potential evaluation model; S42. Use the feature contribution value to evaluate the influence of each input feature on the output of the urban renewal potential evaluation model, and obtain the driving relationship between urban renewal and each input feature based on the influence of each input feature.

Images