Office space diagnosis data processing method

By building an office space diagnostic data model, combining sensor data and user evaluation, design data is generated, the problem of relying on subjective judgments in the existing technology is solved, data-driven office space design is realized, and the scientificity and efficiency of the design is improved.

CN120257411APending Publication Date: 2025-07-04FUTTIERS (SHANGHAI) ARCHITECTURAL DESIGN OFFICE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411263799.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2025-07-04

AI Technical Summary

Technical Problem

The existing office space design and operation methods rely on subjective judgment of experts, lack scientificity and comprehensiveness, sensor data cannot provide comprehensive spatial strategy recommendations, and lack data-driven design models based on past experience.

Method used

By collecting project spatial data, behavioral data and experience data, forming a database, building a data model, conducting univariate correlation analysis and random forest model training, generating design data, and optimizing prediction models through continuous collection of experience data, combining sensor data and user evaluation to establish spatial design adjustment indicators.

Benefits of technology

It reduces the dependence on individual abilities, realizes the scientificity and comprehensiveness of real-time data adjustment and design data, and improves the work efficiency and scientificity of office space diagnostic design.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120257411A_ABST
    Figure CN120257411A_ABST
Patent Text Reader

Abstract

The invention relates to a method for processing office space diagnostic data. The method comprises the following steps: generating a database; processing spatial data; carrying out single-variable correlation analysis; constructing a prediction model; collecting condition data; generating design data; collecting supplementary condition data; and inputting new project data. According to the office space diagnosis data processing method, the data is collected to form the database, the data model is constructed, the design data is generated, the prediction model is improved by continuously collecting the experience data, the function of converting condition data of a new project into space design data is achieved, and through a continuously optimized algorithm, the working efficiency is improved. The dependence of office space diagnostic design on individual ability is reduced; by means of continuous data tracking, real-time adjustment of input data and timely updating of design data, the working efficiency, scientificity and comprehensiveness of diagnostic design are improved, experience driving is replaced with data driving finally, and the decision making process is more scientific.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of architectural design, and particularly to a method for processing office space diagnosis data. Background Art

[0002] Traditional space design research methods and strategy formulation methods rely on the personal experience and consulting means of strategy planners, and their scientificity and rationality remain to be proven.

[0003] At present, some invention achievements in the direction of office space renovation adopt the AHP method to construct indicators, which improves the scientificity of decision-making. However, the limitation of the AHP method is that although the calculation process of weights is an objective mathematical process, the scoring of pairwise comparisons completely depends on the subjective judgment of experts.

[0004] Currently, there are office space management software based on sensor data in the market, which can monitor the operation efficiency through the experience data of sensors. However, these software rarely can further propose suitable space strategies to solve the detected problems. The index dimensions constructed only relying on sensor data are relatively single, and cannot comprehensively consider the multi-dimensional requirements such as the needs of enterprises and employees and the carrier situation.

[0005] Generally speaking, although the existing space design research methods have made certain progress in the direction of dataization, the basic attribute of relying on experts' subjective judgment has not been well improved. At the same time, the current office space management software also cannot provide effective suggestions for space operation. The existing office space design and operation methods do not form a perfect database based on past successful experiences, so a data-driven space design model cannot be established and improved. Summary of the Invention

[0006] To solve the technical problems existing in the prior art, the present invention provides a method for processing office space diagnosis data, so as to form a database of existing office space diagnosis design experiences, construct a data model, and develop into a process that can generate design data through a prediction model and improve the prediction model by continuously collecting experience data.

[0007] To achieve the above object, the technical solution of the present invention is as follows:

[0008] A method for processing office space diagnosis data includes the following steps:

[0009] S1, database generation, collecting project space data, behavior data and experience data and inputting them into the database;

[0010] S2, space data processing, generating optimized space data based on experience data, which will be combined with behavior data for model training in the future;

[0011] S3. Univariate correlation analysis. According to different spatial classifications, linear correlation analysis will be performed on each field, and variables with extremely low correlation will be screened out. The remaining variables will be the features of this spatial classification.

[0012] S4. Perform data preprocessing, and divide the filtered data into a training set and a test set, construct and validate a random forest model.

[0013] S5. Conditional data collection. Collect new customer spatial data and behavioral data, and extract feature labels related to the model.

[0014] S6. Input the collected main conditional data into the prediction model to generate spatial carrier design data.

[0015] S7. Supplementary conditional data collection. Continuously collect experience data during the project operation process.

[0016] S8. Enter the new project spatial data, behavioral data, and experience data into the database.

[0017] As an optimal technical solution, step S2 includes data preprocessing; establishing spatial design adjustment indicators by integrating sensor data and user evaluation data; outputting the adjusted number of spaces and area quantity.

[0018] As an optimal technical solution, establish spatial design adjustment indicators by integrating sensor data and user evaluation data. Based on sensor data and computer spatial simulation technology, simulate the behavioral habits of users in each space, obtain the space utilization rate, capacity utilization rate, and space shortage rate under different spatial data, and obtain the spatial data that makes the quantity satisfaction and area satisfaction closest to the target value.

[0019] As an optimal technical solution, step S4 includes data preprocessing; prediction model generation; model evaluation and model tuning. Data preprocessing includes data cleaning; merging and correlating data; data splitting, and dividing the data set into a 70% training set and a 30% test set.

[0020] As an optimal technical solution, the prediction model generation is to predict whether to set up and area allocation for different spatial classifications. Use the training set to train the random forest model. The steps to judge whether to set up a space include: assuming that the collected main conditional data is X = [x1, x2,..., x p where x i represents the i-th conditional feature, and there are a total of p features; first construct a random forest model to judge whether to set up a space. Assume that there are T1 trees in the random forest model, denoted as For each decision tree T 1,i , randomly extract a subset from the training data to generate a training subset X T1,i ; use the training subset X T1,iTraining decision tree T 1,i Input the test set X into the random forest model. For each decision tree T 1,i a prediction result y will be given 1,i where y 1,i = T 1,i (X). The voting method synthesizes the prediction results of the decision trees to indicate whether to set up the space:

[0021] As a preferred technical solution, the steps for judging the area allocation of the space are as follows: Construct another random forest model, input the test set X into the random forest model, and for each decision tree T 2,i a prediction result y will be given 2,i indicating the area allocation y of the space 2,i = T 2,i (X); The average method is used to synthesize the prediction results of the decision trees to obtain the final prediction result The final space design result includes two parts: whether to set up the space and the area allocation of the space That is

[0022]

[0023] The present invention achieves the following objectives by realizing the conversion from conditional data to design data:

[0024] (1) By continuously optimizing the algorithm, the dependence of the diagnostic design of the office space on individual capabilities is reduced;

[0025] (2) Through continuous data tracking, the input data is adjusted in real time and the design data is updated in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flowchart of the method for processing office space diagnostic data of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0027] The technical solution of the present invention will be further described below in conjunction with the specific embodiments:

[0028] As Figure 1 shown, a method for processing office space diagnostic data includes the following steps:

[0029] S1 Database generation: Enter the known project space data, behavior data, and experience data into the database;

[0030] S1.1 Spatial data entry: As shown in Table 1 below, enter the project data and floor data of known projects into the database, classify and number the specific floor plans in the CAD drawings by module and mark the area, export information such as module numbers and areas and enter them into the database, and enter information such as module classification and number of users in the database;

[0031] Table 1

[0032]

[0033] S1.2 Behavior data entry: As shown in Table 2, enter information such as questionnaires, interviews, and workshops obtained during user research into the database;

[0034] Table 2

[0035]

[0036] S1.2.1 Structured behavior data entry: For structured behavior data, directly enter it into the database, or extract feature tags through methods such as statistical analysis, cluster analysis, and principal component analysis and then enter them into the database;

[0037] S1.2.2 Unstructured behavior data entry: For unstructured behavior data, extract feature tags related to the design theme through text analysis methods such as keyword extraction and then enter them into the database;

[0038] S1.3 Experience data entry: Enter the sensor data and user evaluation data during the actual use of the project into the database;

[0039] S1.3.1 Sensor data entry: Sensor data includes space occupancy data feedback by passive infrared detectors in spaces such as desks and meeting rooms. After cleaning and converting the data, enter it into the database according to unified data specifications and link it to the modules already entered in the spatial data;

[0040] S1.3.2 User evaluation data entry: Enter the questionnaire information obtained during the follow-up research into the database;

[0041] S2 Spatial data processing: As shown in Table 3, collect experience data, generate optimized spatial data based on the experience data, and then combine it with behavior data for model training.;

[0042] S2.1 Data preprocessing: Preprocess the collected sensor data and user evaluation data. The processing steps include identifying and processing duplicate data, missing values, and outliers. After completing the preprocessing, merge the data into a unified dataset according to spatial classification.

[0043] S2.2 Establish evaluation indicators: Establish various spatial design adjustment indicators by integrating sensor data U and user evaluation data S. The specific categories are shown in Table 3:

[0044] Table 3

[0045]

[0046] Construct a linear regression model based on the above data:

[0047] Take the quantity satisfaction as the dependent variable and the space utilization rate as the independent variable;

[0048] Take the area satisfaction as the dependent variable and the capacity utilization rate and space shortage rate as the independent variables;

[0049] Calculate the user satisfaction data model for this project.

[0050] The target value of quantity satisfaction is 4.2 points, and the target value of area satisfaction is 4.2 points. Based on sensor data and computer space simulation technology, simulate the behavior habits of users in each space, obtain the space utilization rate, capacity utilization rate and space shortage rate under different space data, and obtain the space data that makes the quantity satisfaction and area satisfaction closest to the target value.

[0051] S2.3 Spatial optimization and adjustment: Analyze the spatial design adjustment indicators to obtain spatial optimization and adjustment data, including quantity adjustment and area adjustment data. Finally, output the adjusted spatial quantity and area data.

[0052] S3 Univariate correlation analysis: To determine whether the above spatial data and each field of the adjusted behavior data have an impact on the carrier space design, linear correlation analysis will be carried out on each field according to different space classifications, and variables with extremely low correlation will be screened out. The remaining variables are the characteristics of this space classification;

[0053] S4 Prediction model construction: Perform data preprocessing, and divide the filtered data into a training set and a test set, construct and verify a random forest model;

[0054] S4.1 Data preprocessing: Preprocess the data so that the model can better read the data;

[0055] S4.1.1 Data cleaning: Include identifying and processing duplicate data, missing values and outliers; perform one-hot encoding or label encoding on categorical features and convert them into numerical data that the model can process, etc.;

[0056] S4.1.2 Merge and associate data: Merge and associate data in the database, and merge multiple data sources into a unified data set;

[0057] S4.1.3 Data Segmentation: Divide the dataset into a 70% training set and a 30% test set for model training and evaluation;

[0058] S4.2 Prediction Model Generation: To predict whether different spatial classifications are set and area allocations, use the training set to train a random forest model;

[0059] S4.2.1 Determine Whether to Set Space: Assume that the main conditional data collected is

[0060] x = [x1, x2,..., x p

[0061] where xi i represents the i-th conditional feature, and there are a total of p features;

[0062] First, construct a random forest model to determine whether to set space. Assume that there are T1 trees in the random forest model, denoted as

[0063] For each decision tree T 1,i , randomly draw a subset (sampling with replacement) from the training data to generate a training subset X T1,i ; use the training subset X T1,i to train the decision tree T 1,i . Input the test set X into the random forest model, and each decision tree T 1,i will give a prediction result y 1,i , y 1,i = T 1,i (X)

[0064] Use the voting method to comprehensively determine the prediction results of the decision trees, indicating whether to set space:

[0065]

[0066] S4.2.2 Determine the Area Allocation of Space: Construct another random forest model to determine the area allocation of space. The construction process is the same as the above step 6.1.

[0067] Input the test set X into the random forest model, and each decision tree T2,i will give a prediction result y 2,i , indicating the area allocation of space y 2,i = T 2,i (X);

[0068] Use the averaging method to comprehensively determine the prediction results of the decision trees to obtain the final prediction result

[0069]

[0070] ​The final spatial design results include two parts: whether to set up the space and the area allocation of the space That is

[0071] S4.3 Model Evaluation: Predict the test sets of different spatial classifications and evaluate the prediction results; among them, for the problem of whether to set up, accuracy, precision, recall, etc. are used for evaluation; for the area allocation problem, mean squared error, root mean squared error, mean absolute error, R 2 score, etc. are used for evaluation;

[0072] S4.4 Model Tuning: Use methods such as K-fold cross-validation to evaluate the performance of the model, ensure the generalization ability of the model, and its stability on different data sets; combine grid search or random search to find the best combination of hyperparameters;

[0073] S5 Conditional Data Collection: Collect new customer spatial data and behavioral data through various research methods, including the main body situation, main body needs, and spatial conditions, etc.;

[0074] S5.1 Spatial Data Collection: Spatial data includes project and floor data, which are collected through desktop research. The types of spatial data are shown in Table 4:

[0075] Table 4

[0076]

[0077] S5.2 Behavioral Data Collection: Behavioral data includes three sections: organizational mode, department situation and needs, and office mode. The organizational mode information is collected through the combination of desktop research and interviews. The department situation and needs are collected through the combination of department questionnaires and interviews. The user office mode is collected through the combination of questionnaires and workshops.

[0078] S5.3 Feature Label Extraction: For the unstructured data generated in the above data collection process, through text analysis methods such as keyword extraction, extract the feature labels related to the model;

[0079] S6 Design Data Generation: Input the collected main conditional data into the prediction model to generate spatial carrier design data, including whether to set up and area allocation for different spatial classifications, etc.;

[0080] S7 Supplementary Conditional Data Collection: Continuously collect experience data during the project operation process, as shown in Table 5, including spatial usage sensor data and user evaluation data.

[0081] Table 5

[0082]

[0083] New project data entry for S8: Enter the spatial data, behavioral data, and experience data of the new project into the database. For specific steps, see S1.

[0084] Compared with traditional office space design, this method is a design method based on in-depth analysis of the current situation and usage requirements of users. It collects and analyzes relevant information through methods such as desktop research, questionnaires, and interviews, and evaluates the organizational model, users' work patterns, and personal preferences. Based on these evaluation results, specific design suggestions are put forward, and finally an office space design that meets the requirements is provided for the customer.

[0085] The objective of the present invention is to form a database with the existing diagnostic design experience of office spaces, construct a data model, and develop it into a process that can generate design data through a prediction model and improve the prediction model by continuously collecting experience data. This process aims to improve the work efficiency, scientificity, and comprehensiveness of diagnostic design, and ultimately realize replacing experience-driven with data-driven to make the decision-making process more scientific.

[0086] This embodiment is only a further explanation of the present invention, not a limitation of the present invention. After reading this specification, those skilled in the art can make non-creative modifications to this embodiment as needed, but as long as it is within the scope of the claims of the present invention, it is protected by the patent law.

Claims

1. A method for processing office space diagnosis data, characterized in that, It includes the following steps: S1. Database generation: Collect project space data, behavior data, and experience data and input them into the database; S2. Space data processing: Generate optimized space data based on experience data, which will be combined with behavior data for model training later; S3. Univariate correlation analysis: Conduct linear correlation analysis on each field according to different space classifications, screen out variables with extremely low correlations, and the remaining variables are the features of this space classification; S4. Perform data preprocessing, divide the filtered data into a training set and a test set, construct and validate a random forest model; S5. Conditional data collection: Collect new customer space data and behavior data, and extract feature labels related to the model; S6. Input the collected main conditional data into the prediction model to generate space carrier design data; S7. Supplementary conditional data collection: Continuously collect experience data during the project operation; S8. Input the new project space data, behavior data, and experience data into the database.

2. The method for processing office space diagnosis data according to claim 1, characterized in that, The step S2 includes data preprocessing; establishing space design adjustment indicators by integrating sensor data and user evaluation data; outputting the adjusted number of spaces and the number of areas.

3. The processing method of the office space diagnosis data according to claim 2, wherein Establish space design adjustment indicators by integrating sensor data and user evaluation data. Based on sensor data and computer space simulation technology, simulate the behavior habits of users in each space, obtain the space utilization rate, capacity utilization rate, and space shortage rate under different space data, and obtain the space data that makes the quantity satisfaction and area satisfaction closest to the target value.

4. The processing method of the office space diagnosis data according to claim 1, characterized in that The step S4 includes data preprocessing; prediction model generation; model evaluation and model tuning. Data preprocessing includes data cleaning; merging and associating data; data splitting, dividing the data set into a 70% training set and a 30% test set.

5. The method for processing office space diagnosis data according to claim 4, wherein The prediction model is generated to predict whether different spatial classifications are set and area allocations. The steps of using the training set to train the random forest model and determining whether to set the space include: assuming that the main conditional data collected is X = [x1, x2,..., x p , where x i represents the i-th conditional feature, and there are a total of p features; first, construct a random forest model to determine whether to set the space. Assume that there are T1 trees in the random forest model, denoted as For each decision tree T 1,i , randomly extract a subset from the training data to generate a training subset X T1,i ; use the training subset X T1,i to train the decision tree T 1,i , input the test set X into the random forest model, and each decision tree T 1,i will give a prediction result y 1,i , y 1,i = T 1,i (X). The voting method synthesizes the prediction results of the decision trees to indicate whether to set the space:

6. The processing method of the office space diagnosis data according to claim 5, characterized in that, The steps for judging the area allocation of the space are as follows: construct another random forest model, input the test set X into the random forest model, and each decision tree T 2,i will give a prediction result y 2,i , representing the area allocation y of the space 2,i = T 2,i (X); the prediction results of the decision trees are integrated by the averaging method to obtain the final prediction result The final space design result includes two parts: whether to set up the space and the area allocation of the space That is