Adaptive engineering pricing method and device based on data quality driving

Abstract

The disclosure provides a data quality driven adaptive engineering pricing method and device, and relates to the technical field of data processing. The method comprises the following steps: obtaining a plurality of unit prices corresponding to a target pricing unit of a water and electricity engineering, to form a target unit price set; calculating a concentration index, a peak reality index, a skewness direction index and an outlier index of the target unit price set, and performing four-dimensional data quality evaluation to generate a quality grade of the target unit price set; performing a plurality of estimations based on different statistical principles on the target unit price set to generate a plurality of unit price estimation values; determining a fusion weight corresponding to each unit price estimation value according to the quality grade of the target unit price set and a preset fusion weight mapping relationship; and weighting and fusing the plurality of unit price estimation values according to the corresponding fusion weights to obtain a final quotation of the target pricing unit. Based on the technical scheme, the quotation can be quickly made based on the data quality of the unit price set.

Description

Technical Field

[0001] This disclosure relates to the field of data processing technology, and in particular to an adaptive engineering pricing method and apparatus based on data quality driven. Background Technology

[0002] In the process of compiling the maximum bid limit and bid price for hydropower projects, the cost estimator needs to select a value for the limit or bid for each basic pricing unit in the bill of quantities (e.g., concrete pouring, anchor support, steel reinforcement fabrication and installation), and complete the pricing of all items in the bill of quantities within the specified time.

[0003] Currently, the industry commonly uses the method of applying quotas item by item to compile unit prices. For each item in the bill of quantities, the compiler calculates the labor costs, material costs, and machinery costs item by item based on the quota manual and design parameters, and then summarizes them to obtain the unit price. This method yields relatively accurate results and is currently the mainstream technology.

[0004] However, in rapid quotation scenarios for hydropower projects, such as when the deadline for price listing is approaching or when multiple bid sections need to be prepared simultaneously, item-by-item quotation preparation requires extensive searching of quotas, adjustment coefficients, and verification calculations. A single item in the bill of quantities can take tens of minutes, and the entire bid section often requires hours or even days, making it difficult to meet time constraints. Furthermore, some items in the bill of quantities lack suitable quotas, requiring the preparers to supplement quotas by referring to various related unit prices, further consuming time. Therefore, significantly improving the speed of unit price determination is a pressing technical problem in the field of engineering cost estimation.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] To overcome the problems existing in related technologies, this disclosure provides an adaptive engineering pricing method and apparatus based on data quality driving, which can significantly improve the compilation speed of unit price determination.

[0007] According to a first aspect of the present disclosure, an adaptive engineering pricing method based on data quality is provided. The method includes: obtaining multiple unit prices corresponding to a target pricing unit of a hydropower project to form a target unit price set, wherein the target pricing unit refers to a pricing unit in the bill of quantities that requires a separate quote; calculating the concentration index, kurtosis index, skewness direction index, and outlier index of the target unit price set; performing a four-dimensional data quality assessment based on the calculated concentration index, kurtosis index, skewness direction index, and outlier index to generate a quality level for the target unit price set; performing multiple estimates on the target unit price set based on different statistical principles to generate multiple unit price estimates; determining the fusion weight corresponding to each unit price estimate based on the quality level of the target unit price set and a preset fusion weight mapping relationship; and weighting and fusing the multiple unit price estimates according to their corresponding fusion weights to obtain the final quote for the target pricing unit.

[0008] According to a second aspect of the present disclosure, an adaptive engineering pricing device based on data quality is provided. The adaptive engineering pricing device includes: a data acquisition module, an index calculation module, a four-dimensional data quality assessment module, a statistical estimation module, a fusion weight determination module, and a pricing module. The data acquisition module is used to acquire multiple unit prices corresponding to the target pricing unit of a hydropower project, forming a target unit price set, wherein the target pricing unit refers to a pricing unit in the bill of quantities that needs to be quoted separately. The index calculation module is used to calculate the concentration index, kurtosis index, skewness direction index, and outlier index of the target unit price set. The four-dimensional data quality assessment module is used to perform four-dimensional data quality assessment based on the calculated concentration index, kurtosis index, skewness orientation index, and outlier index, generating a quality level for the target unit price set; the statistical estimation module is used to perform multiple estimations of the target unit price set based on different statistical principles, generating multiple unit price estimates; the fusion weight determination module is used to determine the fusion weight corresponding to each unit price estimate based on the quality level of the target unit price set and the preset fusion weight mapping relationship; the pricing module is used to weight and fuse multiple unit price estimates according to their corresponding fusion weights to obtain the final price of the target pricing unit.

[0009] According to a third aspect of the present disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the data quality-driven adaptive engineering pricing method as described in the first aspect.

[0010] According to a fourth aspect of the present disclosure, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the computer-readable instructions, when executed by the processor, implement the data quality-driven adaptive engineering pricing method as described in the first aspect.

[0011] The technical solutions provided by the embodiments of this disclosure may include the following beneficial effects: In this embodiment, when pricing is required, the adaptive engineering pricing system automatically completes all calculations from data acquisition, four-dimensional data quality assessment, multiple statistical estimates to weighted fusion, eliminating the need for manual application of quotas item by item, as well as manual judgment of data quality or selection of statistical measures. This solution achieves a quantitative perception of the reliability of unit price data by generating quality levels, and automatically adjusts the proportion of each estimate in the weighted fusion based on a preset fusion weight mapping relationship. When encountering a target unit price set containing outliers (such as malicious bids or data entry errors), the adaptive engineering pricing system can automatically reduce the impact of outliers; when encountering highly concentrated high-quality data, the adaptive engineering pricing system can maintain accuracy without introducing unnecessary losses. This adaptive capability allows cost estimators to directly use the quotations output by the adaptive engineering pricing system without pre-screening, eliminating, or correcting data, thereby reducing the traditional quota application work of several hours to seconds, significantly improving the speed of preparing rapid quotations for hydropower projects.

[0012] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0013] The accompanying drawings, which are incorporated in and form part of this disclosure, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0014] Figure 1 This is a schematic diagram of an adaptive engineering pricing system architecture based on data quality driven by an embodiment of the present disclosure.

[0015] Figure 2 This is a flowchart illustrating an adaptive engineering pricing method based on data quality driven by an embodiment of the present disclosure.

[0016] Figure 3 This is a schematic diagram of a data quality level distribution provided in an embodiment of the present disclosure.

[0017] Figure 4 This is a schematic diagram of a confidence level distribution provided in an embodiment of the present disclosure.

[0018] Figure 5 This is a schematic diagram comparing a conventional original mean with the final price quoted in this disclosure, provided for an embodiment of the present disclosure.

[0019] Figure 6 This is a hardware structure diagram of a computer device in which the data quality-driven adaptive engineering pricing method provided in this embodiment of the disclosure is located.

[0020] Figure 7This is a schematic diagram of a data quality-driven adaptive engineering pricing device provided in an embodiment of this disclosure. Detailed Implementation

[0021] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0022] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0023] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0024] The embodiments of this disclosure will now be described in detail.

[0025] Figure 1 A schematic diagram of a data quality-driven adaptive engineering pricing system architecture applicable to embodiments of this disclosure is shown. Figure 1 As shown, the adaptive engineering pricing system 100 includes: a data access unit 101, a data quality assessment unit 102, a multi-model parallel computing unit 103, and a user interaction unit 104.

[0026] The data access unit 101 is used to acquire and initially process the original unit prices. The data access unit 101 includes a local file import interface 1011, a manual entry interface 1012, and a database 1013. Users can select unit price files in Excel (spreadsheet software) or CSV (Comma-Separated Values) format locally and upload them to the database 1013 via the local file import interface 1011. The adaptive engineering pricing system automatically parses the file content, identifies the header and data type, imports the structured data into the adaptive engineering pricing system, and performs format verification and outlier warnings. Users can also manually enter unit prices into the database 1013 within the adaptive engineering pricing system via the manual entry interface 1012. The manual unit price entry module within the adaptive engineering pricing system provides a user-friendly interface, supporting users to enter unit prices one by one, batch paste, or edit them online, and performing real-time data format verification to ensure the validity and completeness of the input data. The output of this unit is a structured target unit price set, providing basic data support for subsequent quality assessment and pricing calculations.

[0027] The data quality assessment unit 102 is used to perform multidimensional quality assessment on the input unit price. By calculating four original quality indicators and comparing them with a five-level quality assessment table, the data quality level is determined. The data quality assessment unit 102 includes a four-dimensional data quality assessment unit 1020 and a quality level output unit 1025. The four-dimensional data quality assessment unit 1020 includes a concentration index (CI) solution unit 1021, a peak realism index (PI) solution unit 1022, a skewness direction index (SDI) solution unit 1023, and an outlier index (OI) solution unit 1024. The concentration index solving unit 1021 is used to calculate the concentration index of the target unit price set; the kurtosis index solving unit 1022 is used to calculate the kurtosis index of the target unit price set; the skewness orientation index solving unit 1023 is used to calculate the skewness orientation index of the target unit price set; the outlier index solving unit 1024 is used to calculate the outlier index of the target unit price set; the four-dimensional data quality assessment unit 1020 is used to assess the data quality based on the calculated concentration index, kurtosis index, skewness orientation index, and outlier index; and the quality level output unit 1025 is used to output the assessment results of the four-dimensional data quality assessment unit 1020.

[0028] The multi-model parallel computing unit 103 simultaneously estimates the unit price using three statistical models based on different principles. This parallel computing improves efficiency, and the results output by the data quality assessment unit 102 provide diverse statistical characteristics for subsequent fusion pricing. The multi-model parallel computing unit 103 includes: a robust estimation model construction unit 1031, a density estimation model construction unit 1032, a bootstrap sampling estimation model construction unit 1033, and a final price output unit 1034. The robust estimation model construction unit 1031 constructs a robust estimation model for the target unit price set and calculates robust estimates of the target unit price set based on the constructed robust estimation model. The density estimation model construction unit 1032 constructs a density estimation model for the target unit price set and calculates density estimates of the target unit price set based on the constructed density estimation model. The bootstrap sampling estimation model construction unit 1033 constructs a bootstrap sampling estimation model for the target unit price set and calculates bootstrap sampling unit price estimates of the target unit price set based on the constructed bootstrap sampling estimation model. The final quotation output unit 1034 is used to weight and fuse multiple unit price estimates according to the fusion weights corresponding to the quality level to obtain the final quotation for the target pricing unit.

[0029] User interaction unit 104: mainly responsible for the interaction between the user and the graphical interface.

[0030] Figure 2 This is a flowchart illustrating a data quality-driven adaptive engineering pricing method provided in an embodiment of this disclosure. This method can be used to determine the unit price of the same item in a bid for a hydropower project, either as the maximum bid limit or in the bid price. Figure 2 As shown, the method includes the following steps S201 to S206.

[0031] S201. Obtain multiple unit prices corresponding to the target pricing unit of the hydropower project to form a target unit price set.

[0032] The target pricing unit refers to a pricing unit in the bill of quantities that requires a separate price quote.

[0033] For example: "anchor bolt support / unit", "concrete pouring / cubic meter", or "reinforcing bar fabrication and installation / ton".

[0034] Optionally, multiple unit prices can be historical quotations and / or competitive bid quotations. Historical quotations refer to the transaction or settlement prices of the same construction entity (such as a hydropower development company) for the same pricing unit in similar past projects. For example, multiple unit prices may represent historical quotations for the target pricing unit within a preset timeframe in scenarios such as rapid price limits for similar projects, phased project price references, bid section price limit compilation, and claims and change pricing. Competitive bid quotations refer to bids submitted by multiple bidders for the same pricing unit in the same bidding project. For example, multiple unit prices may represent competitive bid quotations submitted by multiple bidders for the target pricing unit in scenarios such as maximum bid limit verification, multi-section bid price balancing, automatic electronic bidding review, and competitive negotiation / inquiry procurement. Multiple unit prices can also be a mixture of historical quotations and competitive bid quotations. For example, multiple unit prices may represent a target unit price set composed of historical quotations and competitive bid quotations in scenarios such as new project price limit compilation (when quotas are lacking).

[0035] For example, multiple unit prices can originate from historical winning bid prices, price limits for similar projects, or market inquiry records. Data access methods include importing from Excel or CSV files, as well as entering data line by line or batch pasting through an interactive interface. The adaptive engineering pricing system automatically performs format validation when accessing data, such as checking whether the data is a number greater than 0 and whether it is within a preset reasonable range, thereby ensuring the validity and completeness of the input data.

[0036] For example, for the target pricing unit - C30 concrete pouring, 15 unit prices are extracted from the settlement data of multiple completed sections: 550, 554, 650, 675, 645, 350, 465, 485, 556, 540, 612, 604, 663, 587, and 568 (unit: yuan / cubic meter). These 15 values ​​constitute the target unit price set.

[0037] S202. Calculate the concentration index, kurtosis index, skewness orientation index, and outlier index of the target unit price set.

[0038] The concentration index indicates the degree of clustering of the target unit price around the target value.

[0039] The kurtosis index indicates the degree of peaks in the distribution of target unit prices. For example, it indicates the fullness of the core region (the middle 50% of the data).

[0040] The skewness orientation index indicates the degree to which the skewness of the unit price distribution within the target unit price concentration affects data quality. In other words, the skewness orientation reflects the symmetry of the data (left-skewed or right-skewed).

[0041] The outlier index indicates the proportion of outlier unit prices in terms of quantity or impact. In other words, the outlier index reflects the degree to which outliers disturb the overall data.

[0042] It should be noted that the embodiments of this disclosure characterize the quality features of the target unit price set from different dimensions using the four indicators described above. These four indicators together constitute the four-dimensional data quality assessment system of this disclosure, providing a basis for subsequent quality level determination.

[0043] S203. Based on the calculated concentration index, kurtosis index, skewness orientation index, and outlier index, perform a four-dimensional data quality assessment to generate the quality level of the target unit price set.

[0044] Among them, the quality grade is used to indicate the reliability of engineering data.

[0045] In this embodiment of the disclosure, the quality level of the unit price data is divided into five levels.

[0046] S204. Perform multiple estimations on the target unit price set based on different statistical principles to generate multiple unit price estimates.

[0047] It should be noted that using a single statistic, such as the mean or median, often fails to fully reflect the distribution characteristics of the data. By estimating using multiple statistical principles, the distribution characteristics of the acquired unit price data can be considered from multiple perspectives, thereby accurately measuring the quality of the acquired unit price data.

[0048] S205. Based on the quality level of the target unit price set and the preset fusion weight mapping relationship, determine the fusion weight corresponding to each unit price estimate.

[0049] The preset fusion weight mapping relationship can be stored in the weight matrix. When automatically determining the price, the corresponding weight can be obtained by looking up the table.

[0050] S206. The multiple unit price estimates are weighted and merged according to their respective fusion weights to obtain the final price of the target pricing unit.

[0051] Specifically, weighted fusion calculates the sum of the products of each estimate and its corresponding weight. The final quote obtained through weighted fusion retains the accuracy of high-quality data while automatically switching to a robust mode for low-quality data.

[0052] Based on this scheme, when pricing is required, the adaptive engineering pricing system automatically completes all calculations from data acquisition, four-dimensional data quality assessment, multiple statistical estimates to weighted fusion, eliminating the need for manual application of quotas item by item, as well as manual judgment of data quality or selection of statistical measures. This scheme achieves a quantitative perception of the reliability of unit price data by generating quality levels, and automatically adjusts the proportion of each estimate in the weighted fusion based on a preset fusion weight mapping relationship. When encountering a target unit price set containing outliers (such as malicious bids or data entry errors), the adaptive engineering pricing system can automatically reduce the impact of outliers; when encountering highly concentrated high-quality data, the adaptive engineering pricing system can maintain accuracy without introducing unnecessary losses. This adaptive capability allows cost estimators to directly use the quotations output by the adaptive engineering pricing system without pre-screening, eliminating, or correcting data, thereby reducing the traditional quota application work of several hours to seconds, significantly improving the speed of preparing rapid quotations for hydropower projects.

[0053] Optionally, in the data quality-driven adaptive engineering pricing method provided in this embodiment of the present disclosure, after generating the quality level of the target unit price set in S203 above, it may further include the following S207.

[0054] S207. Output the quality level of the target unit price set and the corresponding data quality description and engineering meaning.

[0055] For example, in this embodiment of the disclosure, the quality level of the unit price set is divided into 5 levels, and the data quality description and engineering meaning corresponding to each level are as follows.

[0056] Level 1, data quality description: extremely concentrated, perfectly symmetrical; engineering meaning: highly mature market, highly consistent pricing, can be directly used as the final price.

[0057] Grade A, data quality description: highly concentrated, basically symmetrical; engineering meaning: excellent data quality, high credibility of pricing results.

[0058] Grade B, data quality description: moderately dispersed, slightly skewed; engineering meaning: medium data quality, requiring fine-tuning based on experience.

[0059] Grade C, data quality description: moderately dispersed, significantly skewed; engineering meaning: low data quality, manual review recommended.

[0060] Grade D, data quality description: highly dispersed, severely skewed. Engineering meaning: poor data quality, requiring supplementary data or re-collection.

[0061] For example, for being rated as The target unit price set for anchor bolts; the adaptive engineering pricing system can output a quality description along with the final engineering price: extremely concentrated and perfectly symmetrical; the engineering meaning is: the market is highly mature, the quotations are highly consistent, and it can be directly used as the final price.

[0062] Based on this scheme, the adaptive engineering pricing system automatically outputs the quality level and its corresponding quality description and engineering meaning, allowing cost estimators to quickly understand the reliability of the current data without having to analyze the distribution characteristics of the unit price data themselves. For example, Level 1 indicates that the data can be used directly, while Level 2 (D) indicates that supplementary data or manual review is required. This intuitive output avoids the time-consuming process of manually verifying the original data one by one, enabling cost estimators to quickly identify low-quality items in the list that require special attention. This further shortens the overall quotation preparation time while ensuring the reasonableness of the pricing.

[0063] Optionally, in the data quality-driven adaptive engineering pricing method provided in this embodiment of the present disclosure, the concentration index in S202 can be calculated using S202a, the kurtosis index can be calculated using S202b, the skewness orientation index can be calculated using S202c, and the outlier index can be calculated using S202d.

[0064] S202a. Calculate the concentration index of the target unit price set based on the mean and standard deviation of the target unit price set.

[0065] For example, based on the following formula (1), the concentration index of the target unit price set is calculated according to the mean and standard deviation of the target unit price set.

[0066] ; Formula (1) in, Represents the target unit price set. , This indicates the number of unit prices in the target unit price set. This represents the concentration index of the target unit price set. This represents the mean of the target unit price set. This represents the standard deviation of the target unit price set.

[0067] For example, the value of n ranges from 3 to 500.

[0068] It should be noted that when the concentration index is mapped to the (0, 1] interval, the closer the value is to 1, the more concentrated the unit price and the higher the quality; when all unit prices are completely consistent, the concentration index is 1, and when the unit prices are extremely dispersed, the concentration index approaches 0.

[0069] For example, for the target unit price set (n=15) of C30 concrete pouring mentioned above, the calculated mean is approximately 569.29, and the standard deviation is approximately 82.45. Therefore, the concentration index is approximately 82.45 / 569.29 ≈ 0.145. This value is relatively small, indicating that the unit prices of this group of concrete are relatively dispersed.

[0070] S202b: Calculate the peak realism index based on the 10th, 25th, 75th, and 90th quantiles of the target unit price set.

[0071] It's important to note that the 10th percentile indicates that 10% of the data points in a set are smaller than the given data, representing the lower critical value of the data distribution. The 25th percentile, also known as the lower quartile, indicates that 25% of the data points in a set are smaller than the given data. The 75th percentile, also known as the upper quartile, indicates that 75% of the data points in a set are smaller than the given data. The 90th percentile indicates that 90% of the data points in a set are smaller than the given data, representing the upper critical value of the data distribution.

[0072] For example, based on the following formula (2), the peak realism index is calculated according to the 10th, 25th, 75th and 90th percentiles of the target unit price set.

[0073] ; Formula (2) in, This represents the kurtosis index of the target unit price set. This represents the 25th percentile of the target unit price set. This represents the 75th percentile of the target unit price set. This represents the 10th percentile of the target unit price set. This represents the 90th percentile of the target unit price set.

[0074] It should be noted that the kurtosis index is used to characterize the clustering characteristics (saturation) of data in the core area by comparing the width of the middle 50% of the data with the width of the middle 80% of the data. The closer the kurtosis index is to 1, the more concentrated the target unit price is in the core area, with less tailing or outlier phenomena; the closer the kurtosis index is to 0, the more obvious the long tail or outlier phenomena are in the target unit price.

[0075] For example, for the target unit price set of C30 concrete pouring, the price is calculated after sorting the prices from smallest to largest. ≈495, ≈618, ≈456, If the value is approximately 662, then the peak realism index = (618-495) / (662-456) = 123 / 206 ≈ 0.597. The peak realism index is significantly less than 1, indicating that there is a certain degree of tailing phenomenon in the unit price data of the target unit price set.

[0076] S202c, Calculate the skewness orientation index of the target unit price set based on the 10th, 50th, and 90th quantiles of the target unit price set.

[0077] For example, based on the following formula (3), the skewness orientation index of the target unit price set is calculated according to the 10th, 50th and 90th percentiles of the target unit price set.

[0078] ; Formula (3) in, The index representing the skewness direction of the target unit price set. This represents the 50th percentile (i.e., the median) of the target unit price set.

[0079] Among them, the skewness orientation index can intuitively measure the degree and direction of asymmetry in the data distribution. A skewness orientation index greater than 1 indicates that the distribution is right-skewed (stretching in the high-value area), a skewness orientation index less than 1 indicates that the distribution is left-skewed (stretching in the low-value area), and a skewness orientation index equal to 1 indicates that the distribution is perfectly symmetrical.

[0080] For example, for the concrete unit price set mentioned above, ≈556, =662, =456, then the skewness direction index = (662-556) / (556-456) = 106 / 100 = 1.06, which shows a slight right skewness, indicating that there is a small amount of pulling effect from higher unit prices.

[0081] The skewness orientation index provided in this disclosure has a simple calculation method and strong robustness, making it particularly suitable for engineering applications.

[0082] S202d, calculate the outlier index based on the maximum value, 25th percentile, and 75th percentile of the target unit price set.

[0083] For example, the outlier index of the target unit price set can be calculated based on the following formula (4), according to the 25th percentile, 75th percentile, maximum value and minimum value of the target unit price set.

[0084] ; Formula (4) in, Represents the outlier index of the target unit price set. This represents the maximum value in the target unit price set. This indicates taking the maximum value. This represents the minimum value in the target unit price set.

[0085] The above methods can quantify the degree of disturbance that outliers cause to the overall data, effectively identifying malicious pricing or data entry errors in engineering cost estimates. A higher outlier index indicates that the outlier deviates further from the core data, significantly impacts the mean, and indicates lower data quality. An outlier index exceeding 1.5 is generally considered to indicate the presence of outliers, and exceeding 2.5 is considered a severe outlier.

[0086] For example, for the set of concrete unit prices, the maximum value is 675. =618, =495, then the outlier index = (675-618) / (618-495) = 57 / 123≈0.463, which is less than 1.5, indicating that although there is a slightly high value, it has not yet reached the level of significant anomaly.

[0087] Based on this scheme, the adaptive engineering pricing system automatically calculates quality indicators across four dimensions, eliminating the need for cost estimators to manually analyze the distribution characteristics of unit price data. It quantifies unit price quality from four dimensions: data clustering, core area saturation, symmetry, and outlier disturbances. Compared to traditional methods using only standard deviation or range, this four-dimensional assessment can accurately identify long-tail phenomena (e.g., abnormally high prices in some sections), skewness (e.g., left skew due to overall market downturn), and malicious pricing or data entry errors in hydropower project bidding data. In actual projects, cost estimators can use these indicators to quickly determine the reliability of historical unit prices. When the outlier index exceeds 2.5, the adaptive engineering pricing system automatically issues a warning, indicating the need for manual verification of the original data; when the peak solidity index is too low, it indicates that the data may have a multi-peak distribution and requires careful handling. These automated quality feature outputs allow cost estimators to quickly identify problematic unit price sets without manually checking each piece of original data, reducing the data exploration work that previously required tens of minutes to seconds, significantly improving the speed of data quality assessment.

[0088] Optionally, in the adaptive engineering pricing method based on data quality driven by the present disclosure, the above-mentioned S203 may specifically include S203a1 to S203a3 or S203a1 to S203a4 (the method for determining the dominant indicator).

[0089] S203a1. Use the concentration index of the target unit price set as the main indicator to determine the basic level of the target unit price set.

[0090] It should be noted that the concentration index is the most important indicator reflecting data quality; therefore, the concentration index is used as the primary criterion in the dominant indicator determination method. The basic level is determined solely by the range to which the concentration index falls; the other three indicators do not participate in the determination of the basic level.

[0091] Table 1 is an exemplary table for classifying the quality levels of engineering unit price data according to an embodiment of this disclosure. It includes 5 quality levels, as well as the index range, quality description, and engineering meaning of each quality level.

[0092] Table 1 - Quality Grade Classification Table For example, if the concentration index of a certain set of data is 0.93, falling within the range of 0.90 to 0.95, then the basic level is A.

[0093] S203a2, Using the kurtosis index, skewness direction index and outlier index of the target unit price set as auxiliary indicators, determine whether each auxiliary indicator meets the requirements of the corresponding basic level.

[0094] Specifically, after determining the basic level of the target unit price set, the next step is to determine whether the other three indicators simultaneously meet the standards corresponding to that basic level. Each basic level has specific range requirements for the kurtosis index, skewness direction index, and outlier index. If the actual value of any indicator falls outside the required range for that level, it is considered non-compliant.

[0095] S203a3. When at least two of the auxiliary indicators do not meet the requirements of the basic level, the basic level shall be downgraded by one level to become the final quality level.

[0096] S203a4. When at most one of the auxiliary indicators does not meet the requirements of the basic level, the basic level shall be used as the quality level.

[0097] Understandably, if two or more of the three auxiliary indicators perform poorly, it indicates significant problems with the data in other dimensions, necessitating a reduction in quality level. Once downgraded, the data will not be re-checked; that is, it will only be downgraded once.

[0098] For example, assuming the base grade is Grade A (concentration index 0.93), the required kurtosis index for Grade A is 0.75–0.85, the required skewness index is 0.8–0.9 or 1.1–1.2, and the required outlier index is 1.2–1.5. If the actual kurtosis index is 0.70, the skewness index is 0.75, and the outlier index is 1.8, then all three indicators fail to meet the standard (greater than or equal to 2). Therefore, the grade is downgraded from Grade A to Grade B, and the final quality grade is Grade B.

[0099] Based on this scheme, the adaptive engineering pricing system automatically uses a dominant indicator determination method to convert the four-dimensional quality indicators into explicit quality grades. This method aligns with the cognitive habits of engineering cost estimators—first focusing on whether the data is concentrated, and then checking the degree of deviation in other dimensions. When multiple dimensions deviate simultaneously, the system automatically downgrades the quality grade, avoiding the problems of excessive downgrading due to a single abnormal indicator or ignoring the common issues of multiple dimensions. Cost estimators can directly determine subsequent processing strategies based on the quality grade. Data for Level A and Level B can be directly adopted for pricing decisions; Level B requires fine-tuning based on experience; and Levels C and D require manual review or data supplementation. This automated judgment process eliminates the need for repeated manual comparisons using tables, reducing the judgment time to milliseconds, and ensuring consistent and reproducible results, effectively improving the automation level and decision-making efficiency of the pricing process.

[0100] Optionally, in the data quality-driven adaptive engineering pricing method provided in this disclosure embodiment, the above-mentioned S203 can also adopt a weighted comprehensive scoring method, specifically including S203b1 to S203b3 (weighted comprehensive scoring method).

[0101] S203b1. Normalize the concentration index, kurtosis index, skewness direction index, and outlier index of the target unit price set to generate concentration index score, kurtosis index score, skewness direction index score, and outlier index score.

[0102] It should be noted that, due to the different dimensions and value ranges of the four indicators, they are uniformly mapped to the 0-1 interval for weighted aggregation. The concentration index is already in the 0-1 interval, so its original value can be used directly as the score. The kurtosis index is also in the 0-1 interval and can be used directly. Optionally, the skewness orientation index score and outlier index score can be mapped using a traditional normalization method, or they can be mapped using the normalization method provided in this embodiment.

[0103] For example, in this embodiment of the disclosure, a piecewise linear function is used for mapping based on the following formula (5), and a skewness direction index score is generated based on the skewness direction index.

[0104] ; Formula (5) in, This indicates the skewness direction index score.

[0105] Specifically, the ideal value for the skewness orientation index score is 1.0 (perfect symmetry), and the further it deviates from 1.0, the lower the score. 1.0 represents the ideal target value, indicating perfect data symmetry. In this embodiment, 0.5 represents the allowable half-width of deviation, that is, the allowable... It fluctuates within the range of [0.5, 1.5].

[0106] Based on the following formula (6), an outlier index score is generated according to the outlier index.

[0107] ; Formula (6) Specifically, the ideal value of the outlier index is less than or equal to 1.2 (no significant outliers), and the score decreases linearly after it exceeds 1.2, dropping to 0 when it reaches 2.5. In formula (6) of this embodiment, 1.2 represents the threshold for a full score, 2.5 represents the threshold for a zero score, and 1.3 represents the length of the deduction interval.

[0108] For example, if the skewness orientation index is 1.06, falling between 1 and 1.5, then the skewness orientation index score is (1.5 - 1.06) / 0.5 = 0.88. If the outlier index is 0.463, which is less than 1.2, then the outlier index score is 1.

[0109] S203b2. The overall quality score is obtained by weighting the concentration index score, kurtosis index score, skewness direction index score, and outlier index score with preset weights.

[0110] Understandably, the concentration index reflects the most core quality characteristics of the data and should therefore be given the highest weight; the other three indicators are of similar importance and can be given equal, lower weights.

[0111] For example, based on engineering experience, the concentration index weight is set to 0.4, the kurtosis index weight is set to 0.2, the skewness direction index weight is set to 0.2, and the outlier index weight is set to 0.2.

[0112] Optionally, the overall quality score is equal to the sum of the products of the concentration index score and the concentration index weight, the kurtosis index score and the kurtosis index weight, the skewness direction index score and the skewness direction index weight, and the outlier index score and the outlier index weight.

[0113] For example, the overall quality score can be calculated based on the following formula (7).

[0114] ; Formula (7) in, This represents the overall quality score of the target unit price set. This represents the concentration index coefficient of the target unit price set. The concentration index score represents the set of target unit prices. This represents the peak realism index coefficient. This represents the peak realism index score of the target unit price set. This represents the skewness direction exponent coefficient. The skewness orientation index score represents the target unit price set. Represents the outlier index coefficient. The outlier index score represents the target unit price set.

[0115] For example, for the concrete unit price set mentioned above, the concentration index score = 0.145, the kurtosis index score = 0.597, the skewness direction index score = 0.88, and the outlier index score = 1, then S = 0.4 × 0.145 + 0.2 × 0.597 + 0.2 × 0.88 + 0.2 × 1 = 0.058 + 0.1194 + 0.176 + 0.2 = 0.5534.

[0116] S203b3. Determine the quality level of the target unit price set based on the numerical range of the comprehensive quality score.

[0117] For example, the quality grade classification rule is: S≥0.95 is... Grade A: 0.85≤S<0.95; Grade B: 0.70≤S<0.85; Grade C: 0.55≤S<0.70; Grade D: S<0.55.

[0118] In the example above, S=0.5534, which falls between 0.55 and 0.70. Therefore, the quality level is C, and the quality description is "moderately dispersed, significantly skewed". The engineering meaning is "the data quality is low and manual review is recommended".

[0119] Based on this scheme, the weighted comprehensive scoring method can continuously quantify information from four dimensions, avoiding grade jumps caused by threshold boundaries in the dominant indicator judgment method. For example, when the concentration index is slightly below 0.90 (e.g., 0.89) but other indicators are excellent, the comprehensive score may still reach grade A, avoiding manual review caused by mechanical downgrading; conversely, when the concentration index is 0.91 but other indicators are extremely poor, the comprehensive score will automatically drop to grade B or C, avoiding inflated grades. This refined quality assessment mechanism does not require manual adjustment of boundary thresholds; it is entirely determined automatically by the adaptive engineering pricing system based on continuous scores. This reduces the additional review work caused by grade misjudgment and ensures the accuracy and stability of quality grade output in rapid quotation scenarios.

[0120] Optionally, in the data quality-driven adaptive engineering pricing method provided in this embodiment of the disclosure, the above-mentioned S204 may specifically include S204a, S204b and S204c.

[0121] S204a. Perform robust estimation on the target unit price set to generate robust estimates.

[0122] It should be noted that robust estimation is used to resist the interference of outliers and can solve the problem that traditional methods (such as the mean) fail when the data deviates from the ideal distribution.

[0123] Specifically, S204a can be executed by S204a1 to S204a4 as described below.

[0124] S204a1. Sort the unit prices in the target unit price set in ascending order of value.

[0125] S204a2. Remove the first and last unit prices from the sorted target unit price set, and calculate the mean of the remaining unit prices to obtain the first corrected mean.

[0126] S204a3, Replace the first unit price of the sorted target unit price set with the second unit price, replace the last unit price with the second to last unit price, and calculate the mean of the entire unit price sequence after the replacement to obtain the second corrected mean.

[0127] S204a4. Calculate the robust estimate of the target unit price set by weighting and summing the median, first corrected mean, and second corrected mean of the target unit price set.

[0128] It should be noted that the median, first corrected mean, and second corrected mean of the target unit price set are all insensitive to outliers. By weighting them, the robustness can be further enhanced. Finally, a weighted sum is performed to generate a robust estimate.

[0129] For example, a robust estimate of the target unit price set can be calculated by weighting and summing the median, the first corrected mean, and the second corrected mean of the target unit price set based on the following formula (8).

[0130] ; Formula (8) in, This represents a robust estimate. This represents the median of the target unit price set. This represents the median coefficient. This represents the first corrected mean. This represents the first corrected mean coefficient. This represents the second corrected mean. This represents the second corrected mean coefficient.

[0131] It should be noted that, , and All of these are preset mean weighting coefficients based on engineering experience. For example, they can be set based on experience. , , .

[0132] For example, for the set of concrete unit prices (sorted: 350, 465, 485, 540, 550, 554, 556, 568, 587, 604, 612, 645, 650, 663, 675), the median is calculated to be 556; the first corrected mean (the mean of the 13 numbers after removing 350 and 675) is approximately 567.15; the second corrected mean (the mean of the sequence after replacing 350 with 465 and 675 with 663) is approximately 567.07. Therefore, the robust estimate is 0.5 × 556 + 0.3 × 567.15 + 0.2 × 567.07 = 278 + 170.145 + 113.414 = 561.559 (yuan / cubic meter).

[0133] S204b: Perform density estimation on the target unit price set and generate density estimates.

[0134] Specifically, the price that appears most frequently in the target price set is taken as the mode. When all prices appear uniquely, the median of the target price set is taken as the mode. The mode is then used as the density estimate based on the following formula (9).

[0135] ; Formula (9) in, This represents the density estimate.

[0136] It's important to note that the mode reflects the most prevalent price level in the market. When it's uncertain whether the target price set follows a normal or bimodal distribution, density estimation can capture the most frequent value in the data, i.e., the mode, thus solving the problem of unknown distribution of the target price set data. When all data points appear uniquely, there is no mode; in this case, the median is used instead.

[0137] For example, for the set of anchor bolt unit prices (15 data points, all unique), the mode is the median. After sorting, the median is 200, therefore the density estimate is 200 yuan / bolt. If there are duplicate values ​​in the target unit price set, such as 215 appearing twice in the anchor bolt unit price set, then the mode is 215.

[0138] S204c. Perform a self-sampling estimation on the target unit price set to generate a self-sampling estimate.

[0139] It should be noted that bootstrap estimation is a non-parametric statistical method. With small samples, it is difficult to calculate the precision of statistics (such as standard error and confidence interval). This method generates multiple bootstrap sample sets by resampling the target unit price set multiple times with replacement, calculates the mean of each bootstrap sample set, and finally takes the average of all means as the estimation result. This method can effectively evaluate the stability and confidence interval of sample statistics, and is particularly suitable for small sample scenarios.

[0140] Specifically, the above-mentioned S204c can be executed through the following S204c1 and S204c2.

[0141] S204c1, Execute on the target unit price set Each resampling operation with replacement generates a self-service unit price sample set with the same size as the target unit price set. It is an integer greater than 1.

[0142] Suppose the target unit price set is {11, 13, 12, 14, 15, 18}. Each sampling involves randomly selecting 6 numbers from these 6 numbers (with replacement after each selection). The results of the first 6 resampling operations with replacement are given below.

[0143] =1; The sample set 1 for self-service unit price is: {13, 11, 18, 13, 12, 14}; the mean 1 = 13.50.

[0144] =2; The sample set 2 for self-service unit price is: {14, 14, 11, 15, 18, 12}; the mean 2 = 14.00.

[0145] =3; The sample set of self-service unit prices is: {12, 18, 13, 11, 12, 15}; the mean is 3 = 13.50.

[0146] =4; The sample set of self-service unit prices is: {15, 12, 14, 18, 11, 13}; the mean is 4 = 13.83.

[0147] =5; The sample set of self-service unit prices is: {11, 18, 15, 13, 14, 12}; the mean is 5 = 13.83.

[0148] Repeat this process 1000 times to obtain a sample set of 1000 self-service unit prices.

[0149] S204c2, Calculate the mean of each self-service unit price sample set, and... The average of the values ​​is used as the bootstrap sampling estimate.

[0150] For example, the self-sampling estimate is calculated based on the following formula (10), according to the number of resampling and the mean of the self-sampling unit price sample set.

[0151] ; Formula (10) in, This represents the estimated value from the autosampled sample. Indicates the number of resampling operations. Indicates the first Secondary resampling Indicates the first The mean of the second resampling.

[0152] For example, the number of resampling times is set to 1000, and the sample set generated by each resampling is the same size as the target unit price set data, and then the mean of the sample set is calculated.

[0153] For the set of concrete unit prices (n=15), the first resampling may yield [550, 554, 554, 650, 350, 465, 485, 556, 540, 612, 604, 663, 587, 568, 550], with a mean of approximately 543.2. After repeating this process 1000 times, the average of the 1000 means is calculated, assuming a result of approximately 568.5 yuan / cubic meter.

[0154] Based on this scheme, three estimation models with different principles are used in parallel to address different issues in hydropower unit prices. The robust estimation model integrates the median and multiple corrected means, exhibiting strong resistance to outliers without the need for manual removal of abnormal data. For example, even with an outlier of 9000 in the steel rebar unit price, the robust estimation still provides results close to the normal range. The density estimation model, based on the mode or median, directly captures mainstream market quotations without assuming data distribution patterns, accurately reflecting the most frequently occurring transaction prices in the market. It is suitable for scenarios with multiple duplicate values ​​(such as similar quotations from multiple bidding units). The self-administered resampling estimation model assesses the stability of the estimates through extensive resampling with replacement. For example, 1000 resampling iterations provide stability information for the estimates, making it particularly suitable for small sample scenarios (such as earthwork excavation with a sample size of only 4). It quantifies sampling error and obtains stable estimates without waiting for more data collection. These three models address common issues in hydropower unit prices, such as outliers, multimodal distributions, and small sample fluctuations, complementing each other and overcoming the limitations of traditional methods that rely solely on a single mean. It can achieve an adaptive mechanism that uses density estimation for good data and robust estimation for poor data. Compared with the traditional method of applying quotas item by item, which takes tens of minutes or even hours, the embodiments of this disclosure compress the generation time of multiple estimates to an extremely short time through parallel computing and automated statistical processing, fundamentally meeting the time requirements for rapid quotation in hydropower projects.

[0155] Optionally, in the data quality-driven adaptive engineering pricing method provided in this embodiment of the disclosure, the above-mentioned S205 may specifically include S205a, and furthermore, the above-mentioned S206 may also include the following S206a.

[0156] S205a: Read the weights of robust estimation, density estimation, and autosampled estimation corresponding to the quality level from the preset engineering weight matrix.

[0157] The engineering weight matrix is ​​pre-set based on a large amount of engineering cost data and expert experience, reflecting the adaptive mapping relationship between data quality and model weights.

[0158] For example, the engineering weight matrix for the fusion of estimates is shown in Table 2 below.

[0159] Table 2 - Engineering Weight Matrix for Estimated Value Fusion For example, for the aforementioned unit price data of concrete, the quality grade obtained by the weighted comprehensive scoring method is grade C. Looking up the table, the first weight is 0.5, the second weight is 0.25, and the third weight is 0.25.

[0160] As can be seen from Table 2, the preset fusion weight mapping relationship satisfies the adaptive logic that the higher the data quality, the greater the density estimation weight; and the lower the data quality, the greater the robust estimation weight.

[0161] Specifically, when the data quality level is When the data quality level is Grade A or Grade B, the density estimation model has a higher weight to fully reflect the market's concentrated bids; when the data quality level is Grade C or Grade D, the robust estimation model has a higher weight to resist outlier interference.

[0162] S206a. The robust estimate, density estimate, and autopilot sample estimate are weighted and fused according to their respective fusion weights to obtain the final price of the target pricing unit.

[0163] For example, the final price of the target pricing unit can be calculated based on the following formula (11).

[0164] ; Formula (11) in, This indicates the final price. Indicates the weights corresponding to robust estimates. This represents the weight corresponding to the density estimate. This represents the weight corresponding to the self-sampling estimate.

[0165] Based on this scheme, the adaptive engineering pricing system automatically reads the corresponding weight values ​​from a pre-set weight matrix according to the quality level, realizing adaptive weighting where "the higher the quality, the greater the weight for density estimation; the lower the quality, the greater the weight for robust estimation." This mechanism directly addresses the shortcomings of fixed calculation methods that cannot adapt to changes in data quality when different target unit price sets have varying degrees of dispersion and distribution patterns. In hydropower engineering practice, when the historical unit prices of the same bill of quantities item are highly concentrated (e.g., ... When the data is at level D (e.g., Grade D), the adaptive engineering pricing system significantly increases the density estimation weight to fully reflect the concentrated market price. When the data contains obvious outliers or is highly dispersed (e.g., Grade D), the adaptive engineering pricing system increases the robust estimation weight to 0.6 to resist outlier interference. Since the weight matrix is ​​pre-fixed, the reading operation only requires one table lookup, and the time consumption is negligible. This adaptive mechanism eliminates the need for cost estimators to determine the proportion of each estimate based on experience, avoiding the uncertainty and time consumption of subjective judgment, while ensuring the reasonableness of the price quote under different data quality levels.

[0166] Optionally, in the data quality-driven adaptive engineering pricing method provided in this embodiment of the disclosure, S208 and S209 may be included after S206.

[0167] S208, Based on the data generated during the self-service resampling estimation process. Calculate the confidence interval based on the mean of a sample set of self-service unit prices.

[0168] It should be noted that the data generated by self-service resampling The mean of each sample constitutes an empirical distribution of the estimated value. Based on this distribution, a confidence interval can be calculated to characterize the possible range of fluctuation in the final price. A commonly used confidence level is 95%, at which point the lower limit of the confidence interval is the 2.5 percentile, and the upper limit is the 97.5 percentile.

[0169] For example, sorting the mean of 1000 self-tested concrete data samples in ascending order, taking the 2.5th percentile as 540.2 and the 97.5th percentile as 592.7, yields a 95% confidence interval of [540.2, 592.7] (yuan / cubic meter). This interval indicates that there is a 95% probability that the final quote will fall within this range.

[0170] S209, Output the final quote, quality grade, and confidence interval.

[0171] It is understandable that the adaptive engineering pricing system outputs the final price, quality grade, and confidence interval, which can be used as a reference by those who prepare price limits or quotations for hydropower projects.

[0172] Understandably, the output format can be a PDF (Portable Document Format) report or an Excel spreadsheet for easy archiving and sharing. Cost estimators can judge the stability of pricing based on the width of the confidence interval: a narrower interval indicates more concentrated data and more stable pricing; a wider interval indicates greater data fluctuation and higher pricing risk. Meanwhile, the quality level provides an intuitive indication of reliability.

[0173] For example, the adaptive engineering pricing system outputs: "Final price: 561.91 yuan / cubic meter; Quality grade: C (moderately dispersed, obviously skewed, manual review recommended); 95% confidence interval: [540.2, 592.7] yuan / cubic meter." Based on this scheme, the adaptive engineering pricing system provides quality level and confidence interval along with the final quotation. The confidence interval directly reflects the possible fluctuation range of the quotation, allowing cost engineers to quickly determine the stability of the pricing result without additional sensitivity analysis or risk assessment: a narrower interval indicates more concentrated data, which can be directly adopted; a wider interval indicates greater data fluctuation, requiring supplementary data or manual review. This output method simplifies the work that originally required hours of manual sampling and risk assessment to a few seconds of automatic calculation and display, greatly improving the decision-making efficiency before the price limit is posted online. At the same time, the combined output of quality level and confidence interval allows cost engineers to prioritize low-quality or high-risk items among numerous bill of quantities items, avoiding blindly reviewing each item one by one, further shortening the overall quotation preparation cycle.

[0174] The following four typical engineering cases illustrate the overall technical effects of the embodiments of this disclosure.

[0175] Case 1 (Medium Quality Data): Unit Price of Anchor Bolts.

[0176] With a sample size of 15, the anchor bolt unit price set is: [195, 178, 150, 201, 225, 215, 210, 155, 165, 168, 196, 206, 215, 200, 162]. Through four-dimensional data quality assessment, the anchor bolt unit price set was calculated. =0.8897, =0.7255, =0.4974, =0.4096; The quality level of the anchor bolt unit price set is B, and the fusion weights corresponding to B are, in order, robust estimation weight = 0.4, density estimation weight = 0.3, and autopilot sampling estimation weight = 0.3. The robust estimation model calculates a robust estimate of approximately 192.72 yuan / bolt, the density estimation model calculates a density estimate of approximately 215.0 yuan / bolt, and the autopilot sampling estimation model calculates an autopilot sampling estimate of approximately 189.5 yuan / bolt. The final anchor bolt price = 0.4 × 192.72 + 0.3 × 215 + 0.3 × 189.5 = 198.44 yuan / bolt. The original mean used in the traditional pricing method is 189.4 yuan / bolt. The price in this embodiment is slightly higher than the original mean used in the traditional pricing method. The most frequent anchor bolt unit price set is two prices of 215 yuan / bolt. The mode identified by the density estimation model is 215 yuan / bolt, reflecting the mainstream market transaction price; at the same time The value of 0.8897 indicates a high concentration of anchor bolt unit prices; therefore, assigning a higher weight of 0.53 to the density estimate is reasonable. This disclosed embodiment, while maintaining pricing accuracy, effectively reflects the concentration trend of market prices, making the quoted prices more representative.

[0177] Case 2 (Medium quality data): Concrete unit price.

[0178] With a sample size of 15, the concrete unit price set is: [550, 554, 650, 675, 645, 350, 465, 485, 556, 540, 612, 604, 663, 587, 568]. Through four-dimensional data quality assessment, the concrete unit price set is calculated. =0.8724, =0.4518, =0.9453, =2.3353, the quality grade of the concrete unit price set is C, and the fusion weights corresponding to grade C are as follows: robust estimation weight = 0.5, density estimation weight = 0.25, and autosampled estimation weight = 0.25. The robust estimate calculated by the robust estimation model is approximately 571.35 yuan / cubic meter, the density estimate calculated by the density estimation model is approximately 568.00 yuan / cubic meter, and the autosampled estimation estimate calculated by the autosampled model is approximately 566.44 yuan / cubic meter. Therefore, the final concrete price = 0.5 × 571.35 + 0.25 × 568.00 + 0.25 × 566.44 = 566.93 yuan / cubic meter, while the original mean used in the traditional pricing method is 569.29 yuan / cubic meter. This embodiment of the disclosure identifies the right skewness feature (skewness direction index 1.06) and the moderate outlier index of the data through four-dimensional evaluation, and adaptively increases the robust estimation weight to 0.5, thus obtaining a more reasonable price.

[0179] Case 3 (Low-quality data with outliers): Unit price of steel bars.

[0180] The sample size is 13, and the unit price set for rebar is: [4500, 4600, 4550, 4700, 4690, 4741, 4560, 4500, 4650, 4640, 4560, 9000, 4250], where 9000 is a significant outlier. Through four-dimensional data quality assessment, the unit price set for rebar was calculated. =0.8060、 =0.6014、 =1.3280、 =30.7857, the quality grade of the steel rebar unit price set is determined to be Grade D (highly dispersed, severely skewed, requiring supplementary data or re-collection). The fusion weights corresponding to Grade D are as follows: robust estimation weight = 0.6, density estimation weight = 0.2, and autosampled estimation weight = 0.2. The robust estimation model calculates a robust estimate of 4604.51 yuan / ton, the density estimation model calculates a density estimate of 4530.00 yuan / ton, and the autosampled estimation model calculates a autosampled estimate of 4932.99 yuan / ton. Therefore, the final price of steel rebar = 0.6 × 4604.51 + 0.2 × 4530.00 + 0.2 × 4932.99 = 4655.31 yuan / ton. The original mean used in the traditional pricing method is 4918.54 yuan / ton, while the price in this embodiment is 4655.31 yuan / ton. The reason for this is that the original mean used in the traditional pricing method is significantly inflated by the outlier 9000 (without this outlier, the mean would be approximately 4600 yuan / ton), while the embodiment disclosed in this paper is based on the outlier index. =30.7857 (far exceeding the D-level threshold of 2.5) accurately identified a severe outlier problem in data 9000, automatically classifying the quality level as D and increasing the robust estimation weight to 0.6. The pricing scheme provided in this embodiment effectively resists the interference of outliers, and the pricing result is closer to the concentrated area of ​​the steel rebar unit price set (4500-4741 yuan / ton), rather than the mean distorted by outliers.

[0181] Case 4 (Small sample data): Unit price of earthwork excavation.

[0182] The sample size is 4, and the target unit price set is [15.2, 15.8, 16.1, 15.5]. This embodiment adds a small sample correction mechanism: when the sample size n < 10, the adaptive engineering pricing system automatically adjusts the robust estimation weight to 0.6, and the weights of the density estimation and bootstrap sampling estimation are each adjusted to 0.2 (corresponding to D-level weights). The final earthwork excavation price is 15.65 yuan / cubic meter, consistent with the original mean used in traditional pricing methods, while simultaneously outputting a confidence interval to indicate the risk of an insufficient sample size.

[0183] Case 5 (High-quality data): Unit price of foundation slab concrete.

[0184] Sample size = 20, the unit price set of the foundation slab concrete is: [500, 501, 500, 501, 500, 501, 500, 501, 500, 501, 500, 501, 500, 501, 500, 501, 500, 501]. The calculated values ​​are obtained through four-dimensional data quality assessment. =0.9990, =1.0000, =1.0000, =0.0000, the quality grade of the unit price set of the foundation slab concrete is class, The corresponding fusion weights for each level are as follows: robust estimation weight = 0.2, density estimation weight = 0.5, and autosampled estimation weight = 0.3. The robust estimation model calculates a robust estimate of approximately 500.5 yuan / cubic meter, the density estimation model calculates a density estimate of approximately 500.5 yuan / cubic meter, and the autosampled estimation model calculates an autosampled estimate of approximately 500.5 yuan / cubic meter. The final price of the foundation slab concrete is 0.2 × 500.5 + 0.5 × 500.5 + 0.3 × 500.5 = 500.5 yuan / cubic meter. The original mean used in the traditional pricing method is 500.5 yuan / cubic meter. The pricing provided in this embodiment is highly consistent with the original mean used in the traditional pricing method, verifying the rationality of the pricing method of this disclosure—unnecessary adjustments will not be made when the data quality is excellent. At this time, the density estimation (mode) weight is the highest (0.5), reflecting the dominant position of the mainstream transaction price in the market, and the pricing result is stable and reliable.

[0185] Figure 3 This is a schematic diagram of data quality level distribution provided in an embodiment of the present disclosure, from... Figure 3 As can be seen, the data quality levels corresponding to the comprehensive quality scores of the above five cases are, in order: 0.701-B, 0.643-C, 0.511-D, 0.917-A, and 1.000-A.

[0186] Figure 4 This is a schematic diagram of a confidence level distribution provided in an embodiment of the present disclosure, from... Figure 4 As can be seen, the confidence levels corresponding to the confidence scores of the five cases are as follows: 0.825 - medium confidence, 0.761 - medium confidence, 0.569 - low confidence, 0.827 - medium confidence, and 0.999 - high confidence.

[0187] Figure 5 A schematic diagram comparing a conventional original mean with the final price quoted in this disclosure is provided for embodiments of this disclosure. Figure 5As can be seen from the data, Case 1 uses the original average price of 189.40, while the price using the data quality-driven pricing method proposed in this disclosure is 198.44. Case 2 uses the original average price of 566.93, while the price using the data quality-driven pricing method proposed in this disclosure is 569.29. Case 3 uses the original average price of 4918.54, while the price using the data quality-driven pricing method proposed in the embodiments of this disclosure is 4655.31. Case 4 uses the original average price of 15.65, while the price using the data quality-driven pricing method proposed in this disclosure is 15.65. Case 5 uses the original average price of 500.5, while the price using the data quality-driven pricing method proposed in this disclosure is 500.5.

[0188] Through comparative experiments using five typical test cases, the pricing method provided in this disclosure demonstrates significant technical advantages: In the medium-quality data of Case 1 and Case 2, the prices quoted by the adaptive engineering pricing system (198.44 yuan and 566.93 yuan) showed slight differences from the original averages (189.40 yuan and 569.29 yuan), respectively. In Case 1... =0.4974 indicates a left-skewed data characteristic, with a high weighting for the density estimate (mode 215), suggesting the quoted price is converging towards the mainstream market price; in Case 2 =2.3353 identified outlier interference, and the robust estimation weight was increased to 0.5, effectively mitigating the negative impact of outliers. Through adaptive weight adjustment based on four-dimensional data quality assessment, the adaptive engineering pricing system obtained more reasonable quotes under both types of data with different characteristics.

[0189] In the low-quality data containing outliers in Case 3, the price quoted by the adaptive engineering pricing system (4655.31 yuan) is far better than the original mean (4918.54 yuan). By identifying severe outliers through the outlier index, the weight of the robust estimation model is increased to 0.6, which effectively resists the interference of outliers. The original mean has no resistance to outliers. A single erroneous data can ruin the entire pricing. However, the embodiment of this disclosure identifies the degree of anomaly through OI quantification and automatically switches to a robust model.

[0190] In the small sample data of Case 4, the price quoted by this adaptive engineering pricing system (15.65 yuan) is consistent with the original mean (15.65 yuan). At the same time, the robustness is ensured through the small sample correction mechanism. The original mean also gives a point estimate under small sample conditions, but the user does not know whether it is robust. The pricing method provided by this disclosure identifies the small sample risk. The price quoted by this disclosure is the same as the original mean, which also means that although the adaptive engineering pricing system has a small sample size, it gives the most robust pricing result.

[0191] In the high-quality data of Case 5, the quote from the adaptive engineering pricing system (500.5 yuan) is completely consistent with the original mean (500.5 yuan), verifying the accuracy of the pricing method of this disclosure embodiment under ideal data. Its advantage lies in not destroying good data, fully demonstrating that good data should be used with simple results, and its credibility is extremely high.

[0192] As can be seen from the five cases above, regardless of data quality, the pricing method of this disclosure can quickly output the final price and corresponding credibility indicators without manual quota application or data quality judgment. In the outlier scenario of Case 3, the pricing method of this disclosure automatically identifies outliers and increases robust estimation weights through the outlier index, directly providing a reasonable price. The entire process takes only seconds, avoiding the time-consuming manual screening and removal of outliers. In the small sample scenario of Case 4, the pricing method of this disclosure, through self-service resampling and confidence interval output, enables cost estimators to immediately understand the risks brought by small samples, thereby deciding whether to supplement data without relying on experience and guesswork. This disclosure, through four-dimensional data quality assessment, transforms data quality from implicit assumptions into explicit quantitative indicators, and drives adaptive weighted fusion, thereby significantly improving robustness to complex scenarios such as skewness, outliers, and small samples while maintaining the accuracy of high-quality data. The core assumption of the original mean method is that "all data are equally reliable," but this assumption often does not hold true in real business scenarios. The embodiments disclosed herein maintain accuracy under high-quality data and provide more scientific and robust pricing results than the original mean method even under complex scenarios with medium quality data, outliers, and small samples. In summary, the pricing method provided by the embodiments of this disclosure significantly improves the efficiency and reliability of rapid pricing for hydropower projects.

[0193] Corresponding to the embodiments of the foregoing methods, this disclosure also provides embodiments of the apparatus and the computer equipment on which it is applied.

[0194] Embodiments of the disclosed device can be applied to computer devices, such as servers or terminal devices. The device embodiments can be implemented in software, hardware, or a combination of both. Taking software implementation as an example, a logically defined data quality-driven adaptive engineering pricing device is formed by its processor loading the corresponding computer program instructions from non-volatile memory into memory for execution. From a hardware perspective, such as... Figure 7 The diagram shown is a hardware structure diagram of a computer device housing a data quality-driven adaptive engineering pricing device according to an embodiment of this disclosure. Except for... Figure 6In addition to the processor 610, memory 630, network interface 620, and non-volatile memory 640 shown, the server or electronic device in which the data quality-driven adaptive engineering pricing method is located in the embodiment may also include other hardware depending on the actual function of the computer device, which will not be described in detail here.

[0195] like Figure 7 As shown, Figure 7 An adaptive engineering pricing device based on data quality is provided in this embodiment of the present disclosure. The adaptive engineering pricing device 700 based on data quality includes: a data acquisition module 701, an index calculation module 702, a four-dimensional data quality assessment module 703, a statistical estimation module 704, a fusion weight determination module 705, and a pricing module 706. The data acquisition module 701 is used to acquire multiple unit prices corresponding to the target pricing unit of the hydropower project, forming a target unit price set, wherein the target pricing unit refers to a pricing unit in the bill of quantities that needs to be quoted separately. The index calculation module 702 is used to calculate the concentration index, kurtosis index, and skewness direction of the target unit price set. The system includes: an index and an outlier index; a four-dimensional data quality assessment module 703, used to perform four-dimensional data quality assessment based on the calculated concentration index, kurtosis index, skewness orientation index, and outlier index, generating a quality level for the target unit price set; a statistical estimation module 704, used to perform multiple estimations of the target unit price set based on different statistical principles, generating multiple unit price estimates; a fusion weight determination module 705, used to determine the fusion weight corresponding to each unit price estimate based on the quality level of the target unit price set and a preset fusion weight mapping relationship; and a pricing module 706, used to weight and fuse multiple unit price estimates according to their corresponding fusion weights to obtain the final price of the target pricing unit.

[0196] Optionally, the index calculation module is specifically used to: calculate the concentration index of the target unit price set based on the mean and standard deviation of the target unit price set; calculate the kurtosis index of the target unit price set based on the 10th, 25th, 75th, and 90th quantiles of the target unit price set; calculate the skewness index of the target unit price set based on the 10th, 50th, and 90th quantiles of the target unit price set; and calculate the outlier index of the target unit price set based on the 25th, 75th, maximum, and minimum values ​​of the target unit price set. The concentration index indicates the degree to which unit prices in the target unit price set cluster around the target value; the kurtosis index indicates the peak of the unit price distribution pattern in the target unit price set; the skewness index indicates the degree of influence of the skewness direction of the unit price distribution in the target unit price set on data quality; and the outlier index indicates the proportion of abnormal unit prices in the target unit price set in terms of quantity or influence.

[0197] Optionally, the four-dimensional data quality assessment module is specifically used to: use the concentration index of the target unit price set as the primary indicator to determine the basic level of the target unit price set; use the kurtosis index, skewness direction index, and outlier index of the target unit price set as auxiliary indicators to determine whether each auxiliary indicator meets the requirement range corresponding to the basic level; when at least two of the auxiliary indicators do not meet the requirement range corresponding to the basic level, the basic level is downgraded by one level to the quality level; when at most one of the auxiliary indicators does not meet the requirement range corresponding to the basic level, the basic level is taken as the quality level.

[0198] Optionally, the four-dimensional data quality assessment module is specifically used to: normalize the concentration index, kurtosis index, skewness orientation index, and outlier index to generate concentration index score, kurtosis index score, skewness orientation index score, and outlier index score; perform weighted calculation based on the concentration index score, kurtosis index score, skewness orientation index score, and outlier index score, as well as preset weights, to obtain a comprehensive quality score; and determine the quality level based on the numerical range of the comprehensive quality score.

[0199] Optionally, the statistical estimation module is specifically used for: performing robust estimation on the target unit price set to generate robust estimates; performing density estimation on the target unit price set to generate density estimates; and performing autopilot resampling estimation on the target unit price set to generate autopilot estimates.

[0200] Optionally, the fusion weight determination module is specifically used to: read the weights of robust estimation, density estimation, and auto-sampling estimation corresponding to the quality level from a preset engineering weight matrix.

[0201] Optionally, the statistical estimation module is specifically used for: sorting the unit prices in the target unit price set in ascending order of value; removing the first and last unit prices from the sorted target unit price set, and calculating the mean of the remaining unit prices to obtain the first corrected mean; replacing the first unit price in the sorted target unit price set with the second unit price and the last unit price with the second-to-last unit price, and calculating the mean of the entire unit price sequence after the replacement to obtain the second corrected mean; and performing a weighted summation of the median, the first corrected mean, and the second corrected mean of the target unit price set to generate a robust estimate.

[0202] Optionally, the statistical estimation module is specifically used to: obtain the unit price that appears most frequently in the target unit price set as the mode; when all unit prices appear uniquely, take the median of the target unit price set as the mode, and use the mode as the density estimate.

[0203] Optionally, the statistical estimation module is specifically used for: performing calculations on the target unit price set. Each resampling operation with replacement generates a self-service unit price sample set of the same size as the original target unit price set; among which... The integer is greater than 1; calculate the mean of each self-service unit price sample set to obtain The average value, The average of the values ​​is used as the bootstrap sampling estimate.

[0204] This disclosure provides a data quality-driven adaptive engineering pricing device that automatically completes all calculations from data acquisition, four-dimensional data quality assessment, multiple statistical estimates to weighted fusion when pricing is required. This eliminates the need for manual application of quotas item by item, as well as manual judgment of data quality or selection of statistical measures. By generating quality levels, it achieves a quantitative perception of the reliability of unit price data and automatically adjusts the proportion of each estimate in the weighted fusion based on a preset fusion weight mapping relationship. When encountering a target unit price set containing outliers (such as malicious bids or data entry errors), it can automatically reduce the impact of outliers; when encountering highly concentrated high-quality data, the adaptive engineering pricing system can maintain accuracy without introducing unnecessary losses. This adaptive capability allows cost estimators to directly use the output quotation without pre-screening, eliminating, or correcting data, thereby reducing the traditional quota application work of several hours to seconds, significantly improving the speed of preparing rapid quotations for hydropower projects.

[0205] This disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps as described in the above embodiments of the data quality-driven adaptive engineering pricing method.

[0206] This disclosure also provides a computer device including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the computer-readable instructions, when executed by the processor, implement the steps as described in the above embodiments of the data quality-driven adaptive engineering pricing method.

[0207] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0208] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0209] The foregoing has described specific embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0210] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention applied herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not claimed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0211] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

[0212] The above description is merely a preferred embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A data quality-driven adaptive engineering pricing method, characterized in that, The method includes: Obtain multiple unit prices corresponding to the target pricing unit of the hydropower project to form a target unit price set. The target pricing unit refers to a pricing unit in the bill of quantities that needs to be quoted separately. Calculate the concentration index, kurtosis index, skewness orientation index, and outlier index of the target unit price set; A four-dimensional data quality assessment is performed based on the calculated concentration index, kurtosis index, skewness orientation index, and outlier index to generate the quality level of the target unit price set. Multiple estimates based on different statistical principles are performed on the target unit price set to generate multiple unit price estimates; Based on the quality level of the target unit price set and the preset fusion weight mapping relationship, determine the fusion weight corresponding to each unit price estimate; The multiple unit price estimates are weighted and fused according to their corresponding fusion weights to obtain the final price of the target pricing unit.

2. The method according to claim 1, characterized in that, The calculation of the concentration index, kurtosis index, skewness orientation index, and outlier index of the target unit price set includes: Calculate the concentration index of the target unit price set based on the mean and standard deviation of the target unit price set; Calculate the kurtosis index of the target unit price set based on the 10th, 25th, 75th, and 90th percentiles of the target unit price set; Calculate the skewness orientation index of the target unit price set based on the 10th, 50th, and 90th quantiles of the target unit price set; Calculate the outlier index of the target unit price set based on the 25th percentile, 75th percentile, maximum value, and minimum value of the target unit price set; The concentration index indicates the degree to which the unit prices in the target unit price concentration cluster around the target value; the kurtosis index indicates the peak degree of the distribution pattern of the unit prices in the target unit price concentration; the skewness direction index indicates the degree of influence of the skewness direction of the unit price distribution in the target unit price concentration on data quality; and the outlier index indicates the proportion of abnormal unit prices in the target unit price concentration in terms of quantity or degree of influence.

3. The method according to claim 2, characterized in that, The process of performing a four-dimensional data quality assessment based on the calculated concentration index, kurtosis index, skewness orientation index, and outlier index to generate the quality level of the target unit price set includes: The concentration index of the target unit price set is used as the main indicator to determine the basic level of the target unit price set; The kurtosis index, skewness direction index, and outlier index of the target unit price set are used as auxiliary indicators to determine whether each auxiliary indicator meets the requirement range corresponding to the basic level. When at least two of the auxiliary indicators do not meet the requirements of the basic level, the basic level is downgraded by one level to become the quality level. When at most one of the auxiliary indicators does not meet the requirements of the basic level, the basic level shall be used as the quality level.

4. The method according to claim 2, characterized in that, Based on the calculated concentration index, kurtosis index, skewness orientation index, and outlier index, the quality grade of the target unit price set is generated, including: The concentration index, kurtosis index, skewness orientation index, and outlier index are normalized to generate concentration index score, kurtosis index score, skewness orientation index score, and outlier index score. The overall quality score is obtained by weighting the concentration index score, the kurtosis index score, the skewness direction index score, and the outlier index score with preset weights. The quality level is determined based on the numerical range of the overall quality score.

5. The method according to claim 1, characterized in that, Multiple estimations based on different statistical principles are performed on the target unit price set to generate multiple unit price estimates, including: A robust estimate is generated by performing a robust estimation on the target unit price set; Density estimation is performed on the target unit price set to generate density estimates; Perform a self-sampling resampling estimation on the target unit price set to generate a self-sampling estimate.

6. The method according to claim 5, characterized in that, Based on the quality level and the preset mapping relationship, the fusion weight corresponding to each unit price estimate is determined, including: Read the weights of the robust estimate, density estimate, and autosampled estimate corresponding to the quality level from the preset engineering weight matrix.

7. The method according to claim 5, characterized in that, A robust estimate is performed on the target unit price set to generate robust estimates, including: Sort the unit prices in the target unit price set in ascending order of value; Remove the first and last unit prices from the sorted target unit price set, and calculate the mean of the remaining unit prices to obtain the first corrected mean; The first unit price of the sorted target unit price set is replaced with the second unit price, and the last unit price is replaced with the second to last unit price. The mean of the entire unit price sequence after the replacement is calculated to obtain the second corrected mean. The robust estimate is generated by weighted summation of the median of the target unit price set, the first corrected mean, and the second corrected mean.

8. The method according to claim 5, characterized in that, Density estimation is performed on the target unit price set to generate density estimates, including: The most frequently occurring unit price in the target unit price set is taken as the mode. When all unit prices appear uniquely, the median of the target unit price set is taken as the mode, and the mode is taken as the density estimate.

9. The method according to claim 5, characterized in that, Perform bootstrap resampling estimation on the target unit price set to generate bootstrap sampled estimates, including: Perform on the target unit price set Each resampling operation with replacement generates a self-service unit price sample set with the same data size as the target unit price set; wherein... It is an integer greater than 1; Calculate the mean of each self-service unit price sample set to obtain The average value, the The average of the values ​​is used as the estimated value of the bootstrap sampling.

10. A data quality-driven adaptive engineering pricing device, characterized in that, The adaptive engineering pricing device includes: a data acquisition module, an index calculation module, a four-dimensional data quality assessment module, a statistical estimation module, a fusion weight determination module, and a pricing module; The data acquisition module is used to acquire multiple unit prices corresponding to the target pricing unit of the hydropower project, forming a target unit price set. The target pricing unit refers to a pricing unit in the bill of quantities that needs to be quoted separately. The index calculation module is used to calculate the concentration index, kurtosis index, skewness orientation index, and outlier index of the target unit price set. The four-dimensional data quality assessment module is used to perform four-dimensional data quality assessment based on the calculated concentration index, kurtosis index, skewness orientation index, and outlier index, and generate the quality level of the target unit price set. The statistical estimation module is used to perform multiple estimations on the target unit price set based on different statistical principles, and generate multiple unit price estimates. The fusion weight determination module is used to determine the fusion weight corresponding to each unit price estimate based on the quality level of the target unit price set and the preset fusion weight mapping relationship. The pricing module is used to weight and fuse the multiple unit price estimates according to their corresponding fusion weights to obtain the final price of the target pricing unit.

Images