Method and apparatus for determining improvement path of operation indicators

Abstract

The present invention provides a method and device for determining an improvement path of operation indicators. The method includes: obtaining the indicator parameters of the operation indicators input by a user, processing the indicator parameters using a collective decision-making model of the operation indicators to obtain the labels of the operation indicators, when the labels meet the improvement conditions, determining each improvement path based on the indicator tree of the operation indicators, determining the optimization weight of each improvement path based on the roadbed weight and path abnormality degree of each improvement path, and determining the optimal path among the various improvement paths according to the respective optimization weights. The whole process effectively reduces the degree of manual participation, no longer relies on the work experience of staff to determine the optimal path, reduces the situation of manual errors, improves the accuracy of determining the optimal improvement path, and can quickly determine the optimal improvement path after reducing the degree of manual participation, thereby improving the efficiency of determining the optimal improvement path.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and particularly relates to a method and device for determining an improvement path of operation indicators. Background Art

[0002] A large amount of operation and maintenance data containing indicators is included in a data center. An indicator is a statistical value measuring the overall characteristics of a target and is a numerical indicator capable of characterizing the business status in a certain business activity of an enterprise. An indicator system is to systematically organize scattered single-point indicators with interconnections to form an organic system composed of multiple indicators according to a certain logical relationship and serving a specific purpose.

[0003] In an indicator system, when improving an indicator, usually, staff need to determine the indicator to be improved according to work experience, and then determine the improvement path of the indicator to be improved according to work experience. The whole process has a high degree of manual participation and a high error probability, reducing the accuracy of the determined improvement path. Summary of the Invention

[0004] In view of this, the present invention provides a method and device for determining an improvement path of operation indicators, which use a collective decision-making model to output labels of operation indicators. When the labels meet the improvement conditions, determine the optimization weight of each improvement path of the operation indicator, and determine the optimal improvement path based on the optimization weight. The whole process reduces the degree of manual participation, avoids the situation of manual errors, and improves the accuracy and efficiency of the determined optimal improvement path.

[0005] To achieve the above object, the embodiments of the present invention provide the following technical solutions:

[0006] The first aspect of the present invention discloses a method for determining an improvement path of operation indicators, including:

[0007] Obtain the indicator parameters of the operation indicator input by a user;

[0008] Input the indicator parameters into a pre-constructed collective decision-making model of the operation indicator to obtain the label of the operation indicator under the indicator parameters;

[0009] When the label of the operation indicator meets a preset improvement condition, determine each basic indicator corresponding to the operation indicator;

[0010] Based on the operation indicator and each of the basic indicators, construct an indicator tree with the operation indicator as the root node, and each of the basic indicators corresponds to a sub-node in the indicator tree;

[0011] Determine each path from the root node to each end node in the indicator tree as an improvement path, where the end node has no next-level sub-node;

[0012] For each of the described improvement paths, determine the path weight and path anomaly degree of the improvement path, and determine the optimization weight of the improvement path based on the path weight and path anomaly degree;

[0013] Based on the optimization weights of the respective improvement paths, determine the optimal improvement path among the respective improvement paths.

[0014] A second aspect of the present invention discloses a device for determining an improvement path of an operation index, including:

[0015] A first acquisition unit for acquiring index parameters of an operation index input by a user;

[0016] An input unit for inputting the index parameters into a pre-constructed collective decision-making model of the operation index to obtain a label of the operation index under the index parameters;

[0017] A first determination unit for determining each basic index corresponding to the operation index when the label of the operation index meets a preset improvement condition;

[0018] A first construction unit for constructing an index tree with the operation index as the root node based on the operation index and each of the basic indexes, and each of the basic indexes corresponds to a sub-node in the index tree;

[0019] A second determination unit for determining each path from the root node to each end node in the index tree as an improvement path, where the end node does not have a next-level sub-node;

[0020] A third determination unit for, for each of the improvement paths, determining the path weight and path anomaly degree of the improvement path, and determining the optimization weight of the improvement path based on the path weight and path anomaly degree;

[0021] A fourth determination unit for determining the optimal improvement path among the respective improvement paths based on the optimization weights of the respective improvement paths.

[0022] Compared with the prior art, the present invention has the following advantages:

[0023] The present invention provides a method and device for determining an improvement path of operation indicators. The method includes: obtaining the indicator parameters of the operation indicators input by a user, processing the indicator parameters using a collective decision-making model of the operation indicators to obtain the labels of the operation indicators, when the labels meet the improvement conditions, determining each improvement path based on the indicator tree of the operation indicators, determining the optimization weight of each improvement path based on the roadbed weight and path abnormality degree of each improvement path, and determining the optimal path from each improvement path according to each optimization weight. The whole process effectively reduces the degree of manual participation, no longer relies on the work experience of staff to determine the optimal path, reduces the situation of manual errors, improves the accuracy of determining the optimal improvement path, and can quickly determine the optimal improvement path after reducing the degree of manual participation, improving the efficiency of determining the optimal improvement path. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0025] Figure 1 It is a flowchart of a method for determining an improvement path of operation indicators provided by an embodiment of the present invention;

[0026] Figure 2 It is a schematic structural diagram of an indicator tree provided by an embodiment of the present invention;

[0027] Figure 3 It is a flowchart of a method for determining the path weight and path abnormality degree of an improvement path provided by an embodiment of the present invention;

[0028] Figure 4 It is an example diagram of a baseline coordinate diagram with a double baseline type as the baseline type provided by an embodiment of the present invention;

[0029] Figure 5 It is an example diagram of a baseline coordinate diagram with an upper baseline type as the baseline type provided by an embodiment of the present invention;

[0030] Figure 6 It is an example diagram of a baseline coordinate diagram with a lower baseline type as the baseline type provided by an embodiment of the present invention;

[0031] Figure 7 It is a scenario example diagram of a method for determining an improvement path of operation indicators provided by an embodiment of the present invention;

[0032] Figure 8 It is a flowchart of a method for constructing a collective decision-making model of operation indicators provided by an embodiment of the present invention;

[0033] Figure 9 Flowchart of the method for constructing a decision tree of a data group provided by an embodiment of the present invention;

[0034] Figure 10 Flowchart of a method for obtaining a clustering coefficient and each data interval provided by an embodiment of the present invention;

[0035] Figure 11 Schematic structural diagram of a device for determining an improvement path of an operation index provided by an embodiment of the present invention;

[0036] Figure 12 Schematic structural diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0037] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0038] In this application, the term "including", "comprising" or any other variant thereof is intended to cover a non-exclusive inclusion, such that a process, method, article or device including a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "including a..." does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0039] As can be seen from the background art, in the traditional method of optimizing indicators, the degree of manual participation is high, resulting in a low accuracy rate of the determined improvement path; to solve this problem, the present invention provides a method for determining the improvement path of an indicator, which greatly reduces the degree of manual participation, reduces the probability of human error, and thereby improves the accuracy rate and efficiency of the determined improvement path.

[0040] The present invention can be used in many general or special computing device environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multi-processor devices, distributed computing environments including any of the above devices or equipment, and so on. The execution subject of the present invention can be a processor of a computer or a server.

[0041] Refer to Figure 1, which is a method flowchart of a method for determining an improvement path of an operation index provided by an embodiment of the present invention, and is specifically described as follows:

[0042] S101: Obtain the index parameters of the operation index input by the user.

[0043] There are corresponding business dimension objects and time periods for the index parameters of the operation index; for each operation index, under the same business dimension object, the index parameters in different time periods may be different; similarly, for each operation index, under the same time period, the index parameters in different business dimension objects may be different.

[0044] S102: Input the index parameters into the pre-constructed collective decision-making model of the operation index to obtain the label of the operation index under the index parameters.

[0045] There is a pre-constructed collective decision-making model for each operation index. After obtaining the index parameters of the operation index, input the index parameters into the collective decision-making model of the operation index so that the collective decision-making model processes the index parameters, and thus the label of the operation index under the index parameters can be obtained; further, the label can be used to display the status of the operation index under the index label to the user, and this label can be used to determine whether the operation index needs to be optimized or improved.

[0046] S103: When the label of the operation index meets the preset improvement condition, determine each basic index corresponding to the operation index.

[0047] After obtaining the label of the operation index, determine whether to improve the operation index based on the label of the operation index. Specifically, when the label of the operation index meets the preset improvement condition, it is determined to improve the operation index, and when the label of the operation index does not meet the preset improvement condition, it is determined not to improve the operation index.

[0048] Further, when determining whether the label of the operation index meets the preset improvement condition, it can be judged based on the label value in the label. When the label value is poor, low, bad, etc. and meets the preset label value, it can be determined that the label of the operation index meets the preset improvement condition.

[0049] Further, when determining each basic index corresponding to the operation index, it can be each basic index that has an association relationship with the operation index on the preset business dimension.

[0050] S104: Based on the operation index and each basic index, construct an index tree with the operation index as the root node, and each basic index corresponds to a sub-node in the index tree.

[0051] Construct an index tree with operation metrics as the root node, where each basic metric corresponds to a child node in the index tree. Refer to Figure 2 which is the structural schematic diagram of the index tree provided by the embodiment of the present invention. As shown in the figure, a is the root node of the index tree and also the operation metric; b1, b2, b3, c1, c2, c3, and c4 are all child nodes of the index tree and also the respective basic metrics. In the index tree, the nodes connected to the front end of a node are all the nodes at the next level of that node. Specifically, for example, b1, b2, and b3 are connected to the front end of a, that is, b1, b2, and b3 are all nodes at the next level of a. Therefore, the level of a is higher than that of b1, b2, and b3; c1 and c2 are connected to the front end of b2, that is, c1 and c2 are all nodes at the next level of b2. Therefore, the level of b2 is higher than that of c1 and c2.

[0052] Preferably, there is a preset weight between each node and its next-level node. Taking Figure 2 as an example, the preset weight between a and b1 is W a-b1 , the preset weight between a and b2 is W a-b2 , the preset weight between a and b3 is W a-b3 , the preset weight between b2 and c1 is W a2-c1 , the preset weight between b2 and c2 is W a2-c2 , the preset weight between b3 and c3 is W a3-c3 , the preset weight between b3 and c4 is W b3-c4 .

[0053] S105: Determine the path from the root node to each end node in the index tree as the improvement path.

[0054] In the method provided by the embodiment of the present invention, there are no child nodes at the next level for the end nodes in the index tree. Further, the end nodes are those without child nodes connected to their front ends. Refer to Figure 2 , b1, c1, c2, c3, and c4 are all end nodes in the index tree.

[0055] Determine the path from the root node to each end node in the index tree as the improvement path. Further, taking Figure 2 as an example for illustration, Figure 2 the end nodes in the index tree in Figure 2 are b1, c1, c2, c3, and c4. Then the improvement paths of the index tree in a-b1 include: L a-b2-c1 , representing the path from a to b1; L a-b2-c2 , representing the path from a to c1, and it is necessary to pass through b2 before reaching c1; L a-b3-c3, indicating the path from a to c3, which needs to pass through b3 before reaching c3. Examples will not be given one by one here.

[0056] S106: For each promotion path, determine the path weight and path anomaly degree of the promotion path, and determine the optimization weight of the promotion path based on the path weight and path anomaly degree.

[0057] When improving operation indicators, there will be many promotion paths. In order to make the best use of resources and obtain the best improvement effect, it is necessary to find the best optimization path for the operation indicators among each promotion path. The method provided by the embodiment of the present invention uses the path weight and path anomaly degree of the promotion path to determine the optimization weight of the promotion path, and finally screens out the best promotion path according to the optimization weight of the promotion path.

[0058] It should be noted that in the process of determining the best promotion path, if a certain indicator in the indicator tree lacks data, it is directly prompted that the best promotion path cannot be found. The lack of indicator data means that the value of the indicator in a certain node is empty.

[0059] Furthermore, the path anomaly degree of the promotion path is related to the anomaly degree of the basic indicators in the promotion path. The anomaly degree of the basic indicators is relative to the indicator health degree. If the path weight of the promotion path is very high, but the basic indicators are very healthy, then this promotion path is not the most needed path. Instead, when the path weight of the promotion path is relatively not high, but the anomaly degree of the basic indicators is very high, optimizing this indicator can greatly improve the evaluation of the system. Then this promotion path is very likely to be the optimal promotion path.

[0060] Refer to Figure 3 , which is a flowchart of a method for determining the path weight and path anomaly degree of a promotion path provided by an embodiment of the present invention, and is specifically described as follows:

[0061] S301: Determine the basic indicator corresponding to the end node of the promotion path as the target indicator.

[0062] S302: Obtain the data set of the target indicator, where the data set contains multiple baseline data and multiple samples.

[0063] The process of obtaining the data set of the target indicator is described as follows:

[0064] Extract each sampling data from the historical data of the target indicator within a preset time period;

[0065] Obtain each initial baseline data of the target indicator within the time period;

[0066] Perform normalization processing on each sampling data to obtain each sample data;

[0067] Normalize each initial baseline data to obtain each baseline data;

[0068] Use each sample data and each baseline data to form a data set.

[0069] It should be noted that when extracting each sampling data from historical data, it is related to two dimensions. One is the time dimension, and the extraction method of this dimension is to extract by day. The other is the business dimension, and the extraction method of this dimension is random. The preset time period can be set according to actual needs. Specifically, within the most recent month, the number of sampling data obtained and the number of initial baseline data can be set according to actual needs. Due to the deviation existing in the factual data itself and the initial baseline data often having deviation itself, so in actual scenarios, manually defined initial baseline data is often used.

[0070] Furthermore, the method of normalizing each sampling data and each initial baseline data can be min-max normalization.

[0071] S303: Process each sample data and each baseline data in the data set to obtain a baseline coordinate graph of the target indicator.

[0072] Organize each sample data and each baseline data onto the same coordinate axis, from which a baseline coordinate graph of the target indicator can be obtained. Furthermore, each sample data and each baseline data can be displayed on the coordinate axis.

[0073] S304: Based on the baseline in the baseline coordinate graph, determine the anomaly degree of each sample data.

[0074] The baseline coordinate graph contains a baseline. Preferably, the baseline in the baseline coordinate graph can be composed of each baseline data.

[0075] Furthermore, there are different types of baselines in the baseline coordinate graph. Specifically, when the baseline coordinate graph contains two baselines, namely the upper baseline and the lower baseline, the baseline type in the baseline coordinate graph can be determined as the double-baseline type; when the baseline coordinate graph only contains the upper baseline, the baseline type in the baseline coordinate graph can be determined as the upper-baseline type; when the baseline coordinate graph only contains the lower baseline, the baseline type in the baseline coordinate graph can be determined as the lower-baseline type. Different baseline types determine the anomaly degree of sample data in different ways.

[0076] The specific process of determining the anomaly degree of each sample data based on the baseline in the baseline coordinate graph is as follows:

[0077] Determine the baseline type of the baseline in the baseline coordinate graph;

[0078] Based on the anomaly determination method corresponding to the baseline type, apply the baseline in the baseline coordinate graph to determine each anomaly parameter of each sample data;

[0079] Process each anomaly parameter of each sample data to obtain the anomaly degree of each sample data.

[0080] When the baseline types of the baselines in the baseline coordinate graph are different, the process of determining the anomaly degree of sample data is different. The process of determining the anomaly degree of sample data under different baseline types is described below:

[0081] (1) When the baseline type of the baseline is a double baseline type, the process of determining the anomaly degree of sample data is as follows:

[0082] Refer to Figure 4 , Figure 4 is an example graph of a baseline type of a double baseline in the baseline coordinate graph provided by an embodiment of the present invention; the maximum value of the ordinate in the graph is 1, and the unit of the abscissa is days; baseline up in the graph represents the upper baseline, baseline dowm represents the lower baseline, and X(i) in the graph represents sample point i, where the sample point i corresponds to a piece of sample data, and the value range of i is a positive integer and is specifically related to the number of sample data.

[0083] Furthermore, three offsets associated with X(i) need to be determined, namely h, h1, and h2. h represents the offset of the sample data from the reference value, h1 represents the offset of the sample data from the upper baseline, and h2 represents the offset of the sample data from the lower baseline. Furthermore, the three offsets associated with X(i) are all anomaly parameters of the sample data. Optionally, the reference value can be set as needed or can be a value selected on the upper baseline and the lower baseline.

[0084] Exemplarily: h1 = baseline up - X(i), h2 = X(i) - baseline down ; where baseline up can be any coordinate point on the upper baseline, and baseline dowm can be any coordinate point on the lower baseline.

[0085] Calculate the anomaly degree S i of the sample data using the three offsets associated with X(i). Exemplarily, S i = 1 - min(h1 - h2) / h.

[0086] (2) When the baseline type of the baseline is the upper baseline type, the process of determining the abnormality degree of the sample data is as described below:

[0087] Refer to Figure 5 , Figure 5 , which is an example diagram of the baseline type of the baseline coordinate diagram provided by the embodiment of the present invention being the upper baseline type; the maximum value of the ordinate in the diagram is 1, and the unit of the abscissa is days; baseline up in the diagram represents the upper baseline, and baseline dowm represents the lower baseline. X(i) in the diagram represents sample point i, where this sample point i corresponds to a piece of sample data, and the value range of i is a positive integer, and it is specifically related to the number of sample data.

[0088] Further, two offsets associated with X(i) need to be determined, which are h and h1 respectively. h represents the offset of the sample data from the reference value, and h1 represents the offset of the sample data from the upper baseline. Further, the two offsets associated with X(i) are both abnormal parameters of the sample data. Optionally, the reference value can be set as needed or can be a value selected on the upper baseline.

[0089] Exemplarily: h = baseline up , h1 = baseline up - X(i); where baseline up here can be any coordinate point on the upper baseline.

[0090] Calculate the abnormality degree S of this sample data using the two offsets associated with X(i) i , exemplarily, S i = 1 - h1 / h.

[0091] (3) When the baseline type of the baseline is the lower baseline type, the process of determining the abnormality degree of the sample data is as described below:

[0092] Refer to Figure 6 , Figure 6 , which is an example diagram of the baseline type of the baseline coordinate diagram provided by the embodiment of the present invention being the lower baseline type; the maximum value of the ordinate in the diagram is 1, and the unit of the abscissa is days; baseline up in the diagram represents the upper baseline, and baseline dowm represents the lower baseline. X(i) in the diagram represents sample point i, where this sample point i corresponds to a piece of sample data, and the value range of i is a positive integer, and it is specifically related to the number of sample data.

[0093] Further, two offsets associated with X(i) need to be determined, namely h and h2. h represents the offset between the sample data and the reference value, and h2 represents the offset between the sample data and the lower baseline. Further, the two offsets associated with X(i) are both abnormal parameters of the sample data. Optionally, the reference value can be set as needed or can be a value selected on the lower baseline.

[0094] Exemplarily: h = 1 - baseline down , h2 = X(i) - baseline up ; where baseline down can be any coordinate point in the lower baseline.

[0095] Calculate the abnormality degree S of the sample data using the two offsets associated with X(i) i , exemplarily, S i = 1 - h2 / h.

[0096] Based on the baseline type of the baseline in the baseline coordinate graph, correspondingly select the above three methods for determining the abnormality degree of the sample data to determine the abnormality degree of each sample data.

[0097] S305: Perform a mean operation on each abnormality degree to obtain the path abnormality degree of the improvement path.

[0098] After obtaining the abnormality degree of each sample data, perform an evaluation operation on each abnormality degree, and then obtain the path abnormality degree of the target index. It should be noted that the path abnormality degree of the target index is equivalent to the path abnormality degree of the improvement path.

[0099] Further, the calculation process of the path abnormality degree of the target index is as follows: where m represents the number of sample data, and S i represents the abnormality degree of sample data i.

[0100] S306: Determine the first weight of each group of nodes in the improvement path. Each group of nodes contains two adjacent nodes.

[0101] It should be noted that in the improvement path, two adjacent nodes form a group of nodes. Taking the improvement path L Figure 2 in a-b3-c3 as an example, this path contains two groups of nodes, namely: a - b3, b3 - c3. Each group of nodes has a corresponding preset weight. For example, the preset weight of a - b3 is W a-b3 , and the preset weight of b3 - c3 is W b3-c3 . It should be noted that two adjacent nodes can be the root node and the child node, or two child nodes.

[0102] For each group of nodes, determine the node with a higher level in the group as the target node, determine each child node that is connected to the target node in the index tree and has a lower level than the target node as the target child node, and use the preset weight between each target child node and the target node to obtain the first weight of the group of nodes.

[0103] Continuing with Figure 2 the promotion path L a-b3-c3 as an example for illustration, calculate the first weight M a-b3 of a - b3. It should be noted that the node with a higher level in this group of nodes is a. Therefore, a is the target node. From the Figure 2 index tree, it can be seen that the target child nodes of a are b1, b2, and b3. The specific calculation process is as follows:

[0104] Among them, W a-b1 is the preset weight between a and b1, |W a-b2 | is the preset weight between a and b2, and W a-b3 is the preset weight between a and b3.

[0105] Calculate the first weight M b3-c3 of b3 - c3. It should be noted that the node with a higher level in this group of nodes is b3. Therefore, b3 is the target node. From the Figure 2 index tree, it can be seen that the target child nodes of b3 are c3 and c4. The specific calculation process is as follows:

[0106] Among them, W b3-c3 is the preset weight between b3 and c3, and |W b3-c4 | is the preset weight between b3 and c4.

[0107] S307: Multiply all the first weights to obtain the path weight of the promotion path.

[0108] Continuing the explanation in step S306, the path weight M a-b3-c3 of the promotion path L a-b3-c3 = M a-b3 * M b3-c3 . By multiplying all the first weights, the path weight of the promotion path is obtained.

[0109] For each promotion path, after determining the path abnormality degree and path weight of the promotion path, multiply the path abnormality degree and path weight of the promotion path to obtain the optimization weight of the promotion path. It should be noted that the optimization weight of the promotion path can also be called the path promotion urgency.

[0110] Taking Figure 2 as an example, using the above method, it can be calculated thatFigure 2 The path importance, path anomaly degree, and optimization weight of each promotion path in Figure 2 are shown in Table 1, which is an example of the parameters of the path importance, path anomaly degree, and optimization weight of each promotion path in

[0111] Ascending path Path weight Path anomaly degree Optimization weight <![CDATA[L a-b1 > <![CDATA[M a-b1 > <![CDATA[S a-b1 > <![CDATA[M a-b1 *S a-b1 > <![CDATA[L a-b2-c1 > <![CDATA[M a-b2-c1 > <![CDATA[S a-b2-c1 > <![CDATA[M a-b2-c1 *S a-b2-c1 > <![CDATA[L a-b2-c2 > <![CDATA[M a-b2-c2 > <![CDATA[S a-b2-c2 > <![CDATA[M a-b2-c2 *S a-b2-c2 > <![CDATA[L a-b3-c3 > <![CDATA[M a-b3-c3 > <![CDATA[S a-b3-c3 > <![CDATA[M a-b3-c3 *S a-b3-c3 > <![CDATA[L a-b3-c4 > <![CDATA[M a-b3-c4 > <![CDATA[S a-b3-c4 > <![CDATA[M a-b3-c4 *S a-b3-c4 >

[0112] Table 1

[0113] S107: Determine the optimal promotion path in each promotion path based on the optimization weight of each promotion path.

[0114] Furthermore, the promotion path with the largest optimization weight can be determined as the optimal promotion path of the operation index. It is also possible to select N promotion paths in descending order of the optimization weight, and feedback each selected promotion path to the user. The user can determine the optimal promotion path among the selected promotion paths according to actual needs, where N is a preset value and the value range is a positive integer.

[0115] Refer to Figure 7 , which is a scenario example diagram of the method for determining the promotion path of the operation index provided by the embodiment of the present invention. The business continuity in the figure is the operation index, and the disaster recovery drill execution situation, disaster recovery drill success rate, whether the disaster recovery drill is implemented as planned, disaster recovery construction coverage rate, disaster recovery drill success rate, and application drill scenario coverage rate are all basic indicators. After analyzing the operation index using the method provided by the present invention, the optimal promotion path for improving the operation index is determined as: operation index - emergency drill scenario coverage rate - emergency drill scenario coverage rate. Through this optimal promotion path, the operation index can be improved under the condition of limited resources.

[0116] In the method provided by the embodiment of the present invention, the index parameter of the operation index input by the user is obtained; the index parameter is input into the collective decision-making model of the operation index to obtain the label of the operation index under the index parameter; when the label of the operation index meets the preset improvement condition, each basic index corresponding to the operation index is determined; based on the operation index and each basic index, an index tree with the operation index as the root node is constructed, and each basic index corresponds to a sub-node in the index tree; the paths from the root node to each end node in the index tree are determined as improvement paths, and there are no lower-level sub-nodes at the end nodes; for each improvement path, the path weight and path abnormality of the improvement path are determined, and the optimization weight of the improvement path is determined based on the path weight and path abnormality; based on the optimization weights of each improvement path, the optimal improvement path is determined among each improvement path. By using the collective decision-making model of the operation index to output the label of the operation index, when the label meets the improvement condition, each improvement path is determined, and the optimization weight is determined by using the path weight and path abnormality of the improvement path. Finally, the optimal improvement path is determined among each improvement path by using each optimization weight. The whole process greatly reduces the manual participation degree, reduces the probability of manual errors, obtains the optimal improvement path with very high accuracy, and improves the accuracy and efficiency of determining the optimal improvement path.

[0117] Further, the collective decision-making model of the operation index is described. Refer to Figure 8 , which is the flowchart of the method for constructing the collective decision-making model of the operation index provided by the embodiment of the present invention, and is specifically described as follows:

[0118] S401: Determine each business dimension object of the operation index.

[0119] When determining each business dimension object of the operation index, first determine each business dimension corresponding to the operation index, such as the department dimension, the application system dimension, etc., and then determine each business dimension object corresponding to the operation index under each business dimension. Exemplarily, departments A, B, and C under the business dimension are all business dimension objects, and departments A, B, and C can be determined as each business dimension object of the operation index.

[0120] S402: For each business dimension object, match the business dimension object with each preset time period to obtain multiple dimension parameter groups.

[0121] There are many preset time periods, such as the recent week, the recent two weeks, the recent month, etc. are all preset time periods.

[0122] The dimension parameter group contains a time period and a business dimension object.

[0123] S403: For each group of dimension parameters, obtain the metric values of the operation metrics in the group of dimension parameters.

[0124] It should be noted that the metric values can also be referred to as metric evaluation values, which are values between 0 and 1.

[0125] Exemplarily, when the group of dimension parameters includes business dimension object 1 and time period 1, the metric values corresponding to the operation metrics and business dimension object 1 and time period 1 can be obtained.

[0126] Referring to Table 2, the table of metric values of the operation metrics in each group of dimension parameters provided by the embodiments of the present invention is as follows:

[0127]

[0128] Table 2

[0129] S404: For each time period, aggregate the metric values belonging to the time period to obtain a data group corresponding to the time period.

[0130] Taking Table 2 as an example, the data group corresponding to time period P1 - month contains various metric values, and the various metric values are: 0.12, 0.88, 0.97, 0.02, 0.14, 0.99, 0.46, 0.34.

[0131] S405: For each data group, construct a decision tree corresponding to the data group.

[0132] Use the various metric values in each data group to construct the decision tree of each data group. Refer to Figure 9 , the flowchart of the method for constructing the decision tree of the data group provided by the embodiments of the present invention is specifically described as follows:

[0133] S501: Denoise the data group, and arrange the various metric parameters in the denoised data group in reverse order to obtain a parameter sequence.

[0134] The purpose of denoising the data group is to delete the abnormal metric values in the data group. Further, the isolation forest algorithm is used to filter out the abnormal metric data, and finally the normal metric values are used to construct the decision tree, thereby improving the correctness of the constructed decision tree.

[0135] Further, the various metric parameters in the parameter sequence are sorted from small to large.

[0136] S502: Group the various metric parameters in the parameter sequence to obtain multiple groups of metric parameters, and each group of metric parameters contains two adjacent metric parameters.

[0137] Exemplarily, the parameter sequence contains 4 index parameters, namely index parameter A, index parameter B, index parameter C, and index parameter D. The obtained index parameter groups after grouping are: index parameter group 1: (index parameter A, index parameter B); index parameter group 2: (index parameter B, index parameter C); index parameter group 3: (index parameter C, index parameter D).

[0138] S503: Process each index parameter group to obtain the difference value of each index parameter group.

[0139] Preferably, the first-order difference of each index parameter group can be calculated to obtain the difference value of each parameter group. Exemplarily: Subtract the index parameter with a smaller value in the index parameter group from the index parameter with a larger value to obtain the difference value of the index parameter group.

[0140] S504: Based on the preset number of clusters, process the parameter sequence and each difference value to obtain the clustering coefficient and each data interval.

[0141] Refer to Figure 10 , which is a flowchart of a method for obtaining the clustering coefficient and each data interval provided by an embodiment of the present invention, and is specifically described as follows:

[0142] S601: Determine the first number based on the number of clusters.

[0143] In the method provided by the embodiment of the present invention, the first number is less than the number of clusters. Preferably, the first number is 1 less than the number of clusters; further, the number of clusters ultimately determines the number of data intervals of the decision tree, that is, the number of clusters ultimately determines the number of nodes of the decision tree.

[0144] S602: Select target difference values among each difference value, and the number of target difference values is equal to the first number.

[0145] Exemplarily, when selecting target difference values among each difference value, the difference values can be selected in ascending order of the difference values until the number of selected difference values is equal to the first number, and each selected difference value is determined as the target difference value.

[0146] Preferably, there are various ways to select target difference values from each difference value. For another example: Sort each difference value in ascending order, and select difference values one by one from the difference values at odd sorting positions until the number of selected difference values is equal to the first number, and each selected difference value is determined as the target difference value; The present invention will not elaborate on the method of selecting target difference values.

[0147] S603: Perform operations on each target difference value and each unselected difference value to obtain the clustering coefficient.

[0148] It should be noted that each target difference value can be determined as the difference value between clusters, denoted by R outside ; each unselected difference value is determined as the difference value within a cluster, denoted by R inside ; the clustering coefficient wherein, is the average difference within a cluster, is the average difference between clusters, i = 1, 2, 3,..., n, and n is the number of clusters; is the average of the difference values between each cluster, is the average of the difference values within each cluster.

[0149] S604: For each target difference value, determine the index parameter group to which the target difference value belongs, and perform operations on the index parameters in the index parameter group to obtain a separation value corresponding to the target difference value.

[0150] Perform an average operation on two index parameters in the index parameter group, and the obtained value is the separation value corresponding to the target difference value.

[0151] S605: Apply each separation value to generate each data interval.

[0152] Use each separation value to form a first set, add the value 0 and the value 1 to the first set to obtain an updated first set, arrange the values in the updated first set in ascending order to obtain a first sequence, and generate each data interval based on the first sequence. The data interval is composed of two adjacent values.

[0153] Exemplarily, each separation value is: 0.46, 0.07, 0.7; each data interval obtained based on each separation value is [0, 0.46), [0.46, 0.07), [0.7, 1].

[0154] It should be noted that the number of generated data intervals is equal to the number of clusters.

[0155] S505: For each data interval, determine the number of index parameters in the parameter sequence that are located in the data interval, and determine the parameter proportion of the data interval based on the number of parameters.

[0156] After determining the number of index parameters in the parameter sequence that are located in the data interval, divide the number of parameters by the total number of index parameters in the parameter sequence to obtain the parameter proportion of the data interval.

[0157] S506: Use the clustering coefficient, each data interval, and the parameter proportion of each data interval to construct a decision tree for the data group.

[0158] In the method provided by the embodiments of the present invention, during the process of constructing the decision tree of the data group, the data is screened to remove abnormal data, thereby ensuring the accuracy of the constructed decision tree. Further.

[0159] S406: Based on each decision tree, construct a collective decision model for operation indicators.

[0160] Preferably, all decision trees can be used to construct a collective decision model for operation indicators, or some decision trees can be used to construct a collective decision model for operation indicators. When using some decision trees to construct a collective decision model for operation indicators, select the decision trees used to construct the collective decision model from each decision tree. The selection method can be: determine the weight ratio of each decision tree, and select decision trees in descending order of the weight ratio of each decision tree until the number of selected decision trees is equal to the preset number, and then use the selected decision trees to construct a collective decision model.

[0161] It should be noted that when determining the weight ratio of each decision tree, divide the clustering coefficient of the decision tree by the sum of the clustering coefficients of all decision trees to obtain the weight ratio of the decision tree.

[0162] Preferably, the constructed collective decision model can be refined regularly to output more accurate labels for operation indicators. It should be noted that the definition of the labels is described, and for specific details, refer to Table 3. Table 3 shows the specific content of the definition of the labels provided by the embodiments of the present invention.

[0163]

[0164] Table 3

[0165] It should be noted that the constructed decision tree allows business personnel to selectively perform the following operations: a) interval adjustment; b) pruning and merging; c) decision tree weight adjustment; d) setting label values.

[0166] By applying the method provided by the present invention, the path weight and path anomaly degree of each improvement path of the operation indicator can be calculated, and the optimal improvement strategy for the operation evaluation of the data center can be analyzed. Managers can select a certain operation indicator with a relatively low evaluation in the system visualization view, view the priority improvement suggestions of the subordinate indicators decomposed by this indicator, and render the improvement path with color changes for reference by operation decision-making managers. The present invention uses algorithms such as artificial custom rules and decision trees to analyze the technical operation indicator evaluation data of operation objects, obtain operation object portraits, and update them to the label attributes of operation objects. In the initial stage, according to the importance, characteristic values of one or more technical operation indicators and technical operation experience, label rules are preset, and operation object labels are calculated regularly to improve the efficiency of optimizing indicators, reduce the manual participation degree when optimizing indicators, and improve the accuracy of the determined optimal improvement path.

[0167] Corresponding to the method shown, an apparatus for determining an improvement path of an operation index according to an embodiment of the present invention is provided. This apparatus is used to support the implementation of the method provided by the embodiment of the present invention in practice, and this apparatus can be set in intelligent devices such as computers. Figure 1 Referring to

[0168] FIG. Figure 11 , which is a schematic structural diagram of an apparatus for determining an improvement path of an operation index provided by an embodiment of the present invention, is specifically described as follows:

[0169] The first acquisition unit 701 is configured to acquire index parameters of an operation index input by a user;

[0170] The input unit 702 is configured to input the index parameters into a pre-constructed collective decision-making model of the operation index to obtain a label of the operation index under the index parameters;

[0171] The first determination unit 703 is configured to determine respective basic indexes corresponding to the operation index when the label of the operation index meets a preset improvement condition;

[0172] The first construction unit 704 is configured to construct an index tree with the operation index as the root node based on the operation index and each of the basic indexes, and each of the basic indexes corresponds to a sub-node in the index tree;

[0173] The second determination unit 705 is configured to determine each path from the root node to each end node in the index tree as an improvement path, where the end node does not have a next-level sub-node;

[0174] The third determination unit 706 is configured to, for each of the improvement paths, determine a path weight and a path anomaly degree of the improvement path, and determine an optimization weight of the improvement path based on the path weight and the path anomaly degree;

[0175] The fourth determination unit 707 is configured to determine an optimal improvement path among the improvement paths based on the optimization weights of the improvement paths.

[0176] In the device provided by the embodiment of the present invention, the index parameter of the operation index input by the user is obtained; the index parameter is input into the collective decision-making model of the operation index to obtain the label of the operation index under the index parameter; when the label of the operation index meets the preset improvement condition, each basic index corresponding to the operation index is determined; based on the operation index and each basic index, an index tree with the operation index as the root node is constructed, and each basic index corresponds to a sub-node in the index tree; the path from the root node to each end node in the index tree is determined as the improvement path, and there is no next-level sub-node at the end node; for each improvement path, the path weight and path abnormality of the improvement path are determined, and the optimization weight of the improvement path is determined based on the path weight and path abnormality; based on the optimization weights of each improvement path, the optimal improvement path is determined among each improvement path. By using the collective decision-making model of the operation index to output the label of the operation index, when the label meets the improvement condition, each improvement path is determined, and the optimization weight is determined by using the path weight and path abnormality of the improvement path. Finally, the optimal improvement path is determined among each improvement path by using each optimization weight. The whole process greatly reduces the manual participation degree, reduces the probability of manual errors, obtains an optimal improvement path with very high accuracy, and improves the accuracy and efficiency of determining the optimal improvement path.

[0177] In the device provided by the embodiment of the present invention, it can also be configured as:

[0178] The fifth determination unit is used to determine each business dimension object of the operation index;

[0179] The matching unit is used to match each business dimension object with each preset time period for each business dimension object to obtain multiple dimension parameter groups;

[0180] The second acquisition unit is used to obtain the index value of the operation index in each dimension parameter group for each dimension parameter group;

[0181] The aggregation unit is used to aggregate each index value belonging to the time period for each time period to obtain a data group corresponding to the time period;

[0182] The second construction unit is used to construct a decision tree corresponding to each data group for each data group;

[0183] The third construction unit is used to construct the collective decision-making model of the operation index based on each decision tree.

[0184] In the device provided by the embodiment of the present invention, the second construction unit can be configured as:

[0185] A denoising processing subunit, configured to perform denoising processing on the data group, and arrange each index parameter in the denoised data group in reverse order to obtain a parameter sequence;

[0186] A grouping processing subunit, configured to perform grouping processing on each index parameter in the parameter sequence to obtain a plurality of index parameter groups, where each index parameter group contains two adjacent index parameters;

[0187] A first obtaining subunit, configured to process each index parameter group to obtain a difference value for each index parameter group;

[0188] A second obtaining subunit, configured to process the parameter sequence and each difference value based on a preset number of clusters to obtain a clustering coefficient and each data interval;

[0189] A first determining subunit, configured to, for each data interval, determine the number of index parameters in the parameter sequence that are located in the data interval, and determine the parameter proportion of the data interval based on the number of parameters;

[0190] A constructing subunit, configured to construct a decision tree of the data group by using the clustering coefficient, each data interval, and the parameter proportion of each data interval.

[0191] In the device provided by the embodiment of the present invention, the second obtaining subunit may be configured as:

[0192] A first determining module, configured to determine a first number based on the number of clusters;

[0193] A selecting module, configured to select a target difference value from each difference value, where the number of target difference values is equal to the first number;

[0194] An operation module, configured to perform operations on each target difference value and each unselected difference value to obtain the clustering coefficient;

[0195] A second determining module, configured to, for each target difference value, determine the index parameter group to which the target difference value belongs, and perform operations on the index parameters in the index parameter group to obtain a separation value corresponding to the target difference value;

[0196] A generating module, configured to generate each data interval by applying each separation value.

[0197] In the device provided by the embodiment of the present invention, the third determining unit 706 may be configured as:

[0198] A second determining subunit, configured to determine the basic index corresponding to the end node of the promotion path as the target index;

[0199] A fourth obtaining subunit, configured to obtain a data set of the target metric, where the data set includes a plurality of baseline data and a plurality of samples;

[0200] A fifth obtaining subunit, configured to process each sample data and each baseline data in the data set to obtain a baseline coordinate graph of the target metric;

[0201] A third determining subunit, configured to determine the abnormality degree of each sample data based on the baseline in the baseline coordinate graph;

[0202] A sixth obtaining subunit, configured to perform an averaging operation on each of the abnormality degrees to obtain the path abnormality degree of the improvement path;

[0203] A fourth determining subunit, configured to determine a first weight of each group of nodes in the improvement path, where each group of nodes includes two adjacent nodes;

[0204] A multiplication processing subunit, configured to multiply all the first weights to obtain the path weight of the improvement path.

[0205] In the device provided by an embodiment of the present invention, the third determining subunit may be configured as:

[0206] A third determining module, configured to determine the baseline type of the baseline in the baseline coordinate graph;

[0207] A fourth determining module, configured to apply the baseline in the baseline coordinate graph to determine each abnormality parameter of each sample data based on an abnormality degree determination method corresponding to the baseline type;

[0208] An abnormality parameter processing module, configured to process each abnormality parameter of each sample data to obtain the abnormality degree of each sample data.

[0209] In the device provided by an embodiment of the present invention, the fourth determining subunit may be configured as:

[0210] A fifth determining module, for each group of nodes, determining the node with a higher level in the group of nodes as the target node, determining each child node connected to the target node and with a lower level than the target node in the metric tree as the target child node, and using a preset weight between each target child node and the target node to obtain the first weight of the group of nodes.

[0211] An embodiment of the present invention further provides a storage medium, where the storage medium includes stored instructions, and when the instructions are running, controlling the device where the storage medium is located to execute the method for determining the improvement path of the above operation metric.

[0212] An embodiment of the present invention also provides an electronic device, the structural schematic diagram of which is as follows Figure 12 shown, specifically including a memory 801, and one or more instructions 802, where one or more instructions 802 are stored in the memory 801 and are configured to be executed by one or more processors 803 to execute the method for determining the improvement path of the above-mentioned operation indicators.

[0213] The specific implementation processes and their derivative methods of the above-mentioned various embodiments are all within the protection scope of the present invention.

[0214] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for a system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, reference can be made to the partial description of the method embodiment. The systems and system embodiments described above are only illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative work.

[0215] Those skilled in the art can further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0216] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for determining an improvement path of an operation index, characterized in that, it includes: Obtain the index parameters of the operation index input by the user; Input the index parameters into a pre-constructed collective decision-making model of the operation index to obtain the label of the operation index under the index parameters; When the label of the operation index meets the preset improvement conditions, determine each basic index corresponding to the operation index; Based on the operation index and each of the basic indexes, construct an index tree with the operation index as the root node, and each of the basic indexes corresponds to a sub-node in the index tree; Determine each path from the root node to each end node in the index tree as an improvement path, and there is no next-level sub-node for the end node; For each of the improvement paths, determine the path weight and path abnormality of the improvement path, and determine the optimization weight of the improvement path based on the path weight and path abnormality; Based on the optimization weights of each of the improvement paths, determine the optimal improvement path among each of the improvement paths; Among them, determining the path weight and path abnormality of the improvement path includes: Determine the basic index corresponding to the end node of the improvement path as the target index; Obtain the data set of the target index, and the data set contains multiple baseline data and multiple samples; Process each sample data and each baseline data in the data set to obtain the baseline coordinate map of the target index; Based on the baseline in the baseline coordinate map, determine the abnormality of each sample data; Perform an averaging operation on each of the abnormalities to obtain the path abnormality of the improvement path; Determine the first weight of each group of nodes in the improvement path, and each group of nodes contains two adjacent nodes; Multiply all the first weights to obtain the path weight of the improvement path.

2. The method according to claim 1, characterized in that, The process of constructing a collective decision-making model of an operation index includes: Determine each business dimension object of the operation index; For each of the business dimension objects, match the business dimension object with each preset time period to obtain multiple dimension parameter groups; For each of the dimension parameter groups, obtain the index value of the operation index in the dimension parameter group; For each of the time periods, aggregate each index value belonging to the time period to obtain a data group corresponding to the time period; For each of the data groups, construct a decision tree corresponding to the data group; Based on each of the decision trees, construct a collective decision-making model of the operation index.

3. The method according to claim 2, characterized in that, Constructing the decision tree corresponding to the data group includes: Perform denoising processing on the data group, and arrange each index parameter in the denoised data group in reverse order to obtain a parameter sequence; Perform grouping processing on each index parameter in the parameter sequence to obtain multiple index parameter groups, and each index parameter group contains two adjacent index parameters; Process each of the index parameter groups to obtain the difference value of each of the index parameter groups; Based on a preset number of clusters, process the parameter sequence and each of the difference values to obtain a clustering coefficient and each data interval; For each of the data intervals, determine the number of parameter values of the index parameters in the parameter sequence that are within the data interval, and determine the parameter proportion of the data interval based on the number of parameter values; Use the clustering coefficient, each of the data intervals, and the parameter proportion of each data interval to construct a decision tree for the data group.

4. The method according to claim 3, wherein, the processing the parameter sequence and each of the difference values based on a preset number of clusters to obtain a clustering coefficient and each data interval includes: Based on the number of clusters, determine a first number; Select target difference values from each of the difference values, the number of the target difference values being equal to the first number; Perform operations on each of the target difference values and the unselected difference values to obtain the clustering coefficient; For each of the target difference values, determine the index parameter group to which the target difference value belongs, and perform operations on the index parameters in the index parameter group to obtain a separation value corresponding to the target difference value; Apply each of the separation values to generate each data interval.

5. The method according to claim 1, wherein, the determining the abnormality degree of each sample data based on the baseline in the baseline coordinate diagram includes: Determine the baseline type of the baseline in the baseline coordinate diagram; Based on the abnormality degree determination method corresponding to the baseline type, apply the baseline in the baseline coordinate diagram to determine each abnormality parameter of each sample data; Process each abnormality parameter of each sample data to obtain the abnormality degree of each sample data.

6. The method according to claim 1, wherein, the determining the first weight of each group of nodes in the improvement path includes: For each group of nodes, determine the node with a higher level in the group as the target node, determine each child node connected to the target node in the index tree and with a lower level than the target node as the target child node, and use the preset weight between each target child node and the target node to obtain the first weight of the group of nodes.

7. An apparatus for determining an improvement path of an operation index, wherein, comprises: A first acquisition unit, configured to acquire the index parameters of the operation index input by a user; An input unit, configured to input the index parameters into a pre-constructed collective decision model of the operation index to obtain a label of the operation index under the index parameters; A first determination unit, configured to determine each basic index corresponding to the operation index when the label of the operation index meets a preset improvement condition; A first construction unit, configured to construct an index tree with the operation index as the root node based on the operation index and each of the basic indexes, each basic index corresponding to a child node in the index tree; A second determination unit, configured to determine each path from the root node to each end node in the index tree as an improvement path, where the end node has no next-level child node; A third determination unit, configured to determine, for each of the lifting paths, a path weight and a path abnormality degree of the lifting path, and determine an optimization weight of the lifting path based on the path weight and the path abnormality degree; A fourth determination unit, configured to determine an optimal lifting path from among the lifting paths based on the optimization weights of the respective lifting paths; Wherein, the third determination unit is further configured to determine a basic index corresponding to an end node of the lifting path as a target index; obtain a data set of the target index, the data set including a plurality of baseline data and a plurality of samples; process each sample data and each baseline data in the data set to obtain a baseline coordinate graph of the target index; determine an abnormality degree of each sample data based on a baseline in the baseline coordinate graph; perform an averaging operation on the respective abnormality degrees to obtain a path abnormality degree of the lifting path; determine a first weight of each group of nodes in the lifting path, where each group of nodes includes two adjacent nodes; and multiply all the first weights to obtain a path weight of the lifting path.

8. The apparatus according to claim 7, wherein, it further includes: A fifth determination unit, configured to determine respective business dimension objects of the operation index; A matching unit, configured to match, for each of the business dimension objects, the business dimension object with each preset time period to obtain a plurality of dimension parameter groups; A second obtaining unit, configured to obtain, for each of the dimension parameter groups, an index value of the operation index in the dimension parameter group; An aggregation unit, configured to aggregate, for each of the time periods, the respective index values belonging to the time period to obtain a data group corresponding to the time period; A second construction unit, configured to construct, for each of the data groups, a decision tree corresponding to the data group; A third construction unit, configured to construct a collective decision model of the operation index based on the respective decision trees.

9. The apparatus according to claim 8, wherein, the second construction unit includes: A denoising processing subunit, configured to perform denoising processing on the data group, and arrange each index parameter in the denoised data group in a reverse order to obtain a parameter sequence; A grouping processing subunit, configured to perform grouping processing on each index parameter in the parameter sequence to obtain a plurality of index parameter groups, where each index parameter group includes two adjacent index parameters; A first obtaining subunit, configured to process each of the index parameter groups to obtain a difference value of each of the index parameter groups; A second obtaining subunit, configured to process the parameter sequence and the respective difference values based on a preset number of clusters to obtain a clustering coefficient and respective data intervals; A first determination subunit, configured to determine, for each of the data intervals, a number of index parameters in the parameter sequence located in the data interval, and determine a parameter proportion of the data interval based on the number of parameters; A construction subunit, configured to construct a decision tree of the data group using the clustering coefficient, the respective data intervals, and the parameter proportion of each of the data intervals.

Images