Data analysis method and device, storage medium and program product

Abstract

The invention provides a data analysis method and device, a storage medium and a program product. The method comprises: based on a to-be-analyzed time period of a specified service, determining a first time period and a second time period in historical reference time periods of the to-be-analyzed time period, the historical reference time periods including a plurality of periods, the first time period including an in-phase time period of the to-be-analyzed time period in each period, and the second time period including an in-phase time period; the second time period is other time periods except the first time period in the historical reference time periods; obtaining time sequence service data of the specified service in a first time period according to a first obtaining mode to obtain first time sequence service data, and obtaining time sequence service data of the specified service in a second time period according to a second obtaining mode to obtain second time sequence service data, the data acquisition interval of the first acquisition mode is smaller than the data acquisition interval of the second acquisition mode; and performing data analysis based on the obtained time sequence business data to obtain an analysis result corresponding to the to-be-analyzed time period.

Description

Technical Field

[0001] One or more embodiments of the present specification relate to the field of data processing technology, and in particular, to a data analysis method, device, storage medium, and program product. Background Art

[0002] Data analysis refers to the inspection, cleaning, conversion or modeling of collected data to discover useful information. At present, data analysis is applied in many fields. Taking the application of data analysis in intelligent operation and maintenance as an example, data analysis is illustrated. Intelligent operation and maintenance can obtain and analyze the data generated by IT infrastructure in real time and automatically identify anomalies. Moreover, after identifying the anomaly, anomaly detection is performed on the anomaly and the data points before the anomaly, and the cause of the problem is analyzed based on the anomaly detection results.

[0003] Among them, when analyzing the data generated by IT infrastructure and automatically identifying anomalies, a single-point anomaly detection method is usually used: starting from the detection point, a data point is taken every 10 data points in the historical direction. After sampling multiple days of time series business data, it is predicted whether there is an anomaly at the detection point based on the sampled multi-day time series business data.

[0004] However, the above method is only applicable to anomaly detection of a single point. When batch anomaly detection is required for the anomaly point and the data points in a period of time before the anomaly point, the above method cannot be implemented directly. The anomaly point and the data points of each minute in a period of time before the anomaly point can only be used as detection points in turn and repeated multiple times. At this time, the detection speed of anomaly detection cannot be improved. Summary of the invention

[0005] In view of this, one or more embodiments of this specification provide a data analysis method, device, storage medium and program product, and the technical solution is as follows:

[0006] In a first aspect, this specification provides a data analysis method, comprising:

[0007] Based on the time period to be analyzed of the designated service, determine a first time period and a second time period in a historical reference time period of the time period to be analyzed, wherein the historical reference time period includes a plurality of cycles, the first time period includes a time period in phase with the time period to be analyzed in each cycle, and the second time period is a time period other than the first time period in the historical reference time period;

[0008] Acquire the time series service data of the designated service in the first time period in a first acquisition manner to obtain first time series service data, and acquire the time series service data of the designated service in the second time period in a second acquisition manner to obtain second time series service data, wherein the data acquisition interval of the first acquisition manner is smaller than the data acquisition interval of the second acquisition manner;

[0009] Data analysis is performed based on the acquired time series business data to obtain analysis results corresponding to the time period to be analyzed.

[0010] In a second aspect, the present specification also provides a computer device, comprising: a memory and a processor; the memory stores a computer program that can be executed by the processor; when the processor runs the computer program, it executes the data analysis method described in the first aspect above.

[0011] In a third aspect, the present specification also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the data analysis method as described in the first aspect above is executed.

[0012] In a fourth aspect, the present specification also provides a computer program product, including a computer program, which, when executed by a processor, executes the data analysis method as described in the first aspect above.

[0013] In summary, this specification obtains the time series business data of the first time period according to the first acquisition method, and obtains the time series business data of the second time period according to the second acquisition method, so as to obtain the time series business data used for data analysis. Since the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method, and the first time period is the same phase time period as the time period to be analyzed, more time series business data closely related to the time period to be analyzed is retained, so it can be used to perform batch analysis on multiple moments within the time period to be analyzed, and ensure the accuracy of the batch analysis, without the need to separately and repeatedly obtain and analyze the time series business data for multiple moments, thereby improving the overall data analysis efficiency. Therefore, by applying the data analysis method provided in this specification to anomaly detection scenarios, batch anomaly detection can be achieved at a faster detection speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a schematic diagram of a system architecture provided by an exemplary embodiment;

[0015] Figure 2 is a flowchart of a data analysis method provided by an exemplary embodiment;

[0016] Figure 3is a flowchart of a data analysis method provided by an exemplary embodiment;

[0017] Figure 4 is a flowchart of a data analysis method provided by an exemplary embodiment;

[0018] Figure 5 It is a schematic diagram comparing a single-point anomaly detection method and a batch anomaly detection method provided by an exemplary embodiment;

[0019] Figure 6 is a structural schematic diagram of a computer device provided by an exemplary embodiment;

[0020] Figure 7 It is a structural schematic diagram of a data analysis device provided by an exemplary embodiment. DETAILED DESCRIPTION

[0021] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this manual are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0022] Data analysis refers to the inspection, cleaning, conversion or modeling of collected data to discover useful information. At present, data analysis is applied in many fields. Taking the application of data analysis in intelligent operation and maintenance as an example, data analysis is illustrated.

[0023] Intelligent operation and maintenance is a method that applies artificial intelligence and machine learning technologies to IT (Information Technology) operation and maintenance, aiming to improve the efficiency and accuracy of IT operation and maintenance through automation and intelligent means. Intelligent operation and maintenance can acquire and analyze the data generated by IT infrastructure in real time, automatically identify abnormal points; and after identifying the abnormal point, perform anomaly detection on the abnormal point and the data points before the abnormal point for a period of time, and analyze the cause of the problem based on the anomaly detection results.

[0024] In intelligent operation and maintenance, when analyzing the data generated by IT infrastructure and automatically identifying anomalies, multi-day and minute-level time series business data is usually used for anomaly detection. Therefore, the data length of anomaly detection is very long, the detection speed is very slow, and the requirements for anomaly detection algorithms are also very high. To this end, sampling is currently used to reduce the data length and increase the detection speed. For example, starting from the detection point, a data point is taken every 10 data points in the historical direction. After sampling multi-day time series business data, it is predicted whether there is an anomaly at the detection point based on the sampled multi-day time series business data.

[0025] However, the above method is only applicable to anomaly detection of a single point. When batch anomaly detection is required for the anomaly point and the data points in a period of time before the anomaly point, the above method cannot be implemented directly. The anomaly point and the data points of each minute in a period of time before the anomaly point can only be used as detection points in turn and repeated multiple times. At this time, the detection speed of anomaly detection cannot be improved.

[0026] Based on this, this specification provides a technical solution that can obtain time-series business data of different time periods by using different data collection intervals. On the basis of ensuring that analysis results of more moments in the time period to be analyzed can be obtained, the data length of anomaly detection can be reduced, thereby realizing batch anomaly analysis at a faster detection speed.

[0027] During implementation, this specification first determines the first time period and the second time period in the historical reference time period of the time period to be analyzed based on the time period to be analyzed of the specified service, wherein the historical reference time period includes multiple cycles, the first time period includes the same phase time period of the time period to be analyzed in each cycle, and the second time period is the other time period in the historical reference time period except the first time period. Then, this specification obtains the time series service data of the specified service in the first time period according to the first acquisition method to obtain the first time series service data, and obtains the time series service data of the specified service in the second time period according to the second acquisition method to obtain the second time series service data. Among them, the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method. Finally, this specification can perform data analysis based on the acquired time series service data to obtain the analysis results corresponding to the time period to be analyzed.

[0028] In the above technical scheme, this specification obtains the time series business data of the first time period according to the first acquisition method, and obtains the time series business data of the second time period according to the second acquisition method, so as to obtain the time series business data used for data analysis. Since the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method, and the first time period is the same phase time period as the time period to be analyzed, more time series business data closely related to the time period to be analyzed are retained, so it can be used to perform batch analysis on multiple moments within the time period to be analyzed, and ensure the accuracy of batch analysis, without the need to obtain and analyze the time series business data separately and repeatedly for multiple moments, thereby improving the overall data analysis efficiency. Therefore, by applying the data analysis method provided in this specification to the anomaly detection scenario, batch anomaly detection can be achieved at a faster detection speed.

[0029] The data analysis method provided in this specification can be performed by a computer device. In one embodiment shown, the computer device is a terminal, which can be a mobile phone, a smart wearable device, a tablet computer, a laptop computer, a desktop computer, and a car computer, etc. This specification does not specifically limit this. In one embodiment shown, the computer device is a server, which can be a server, or a server cluster composed of several servers, or a cloud computing service center. Of course, the server can also include other functional servers to provide more comprehensive and diversified services.

[0030] In an illustrated embodiment, the data analysis method may also be executed by multiple computer devices. Figure 1 , Figure 1 This is a schematic diagram of a system architecture provided by an exemplary embodiment. One or more embodiments provided in this specification may be Figure 1 The system architecture shown in the figure or a similar system architecture is specifically implemented. Figure 1 As shown, the system may include a computer device 100a and at least one computer device 100b. In an illustrated embodiment, the computer device 100a and the computer device 100b may establish a communication connection via a wireless network, which is not specifically limited in this specification.

[0031] In an illustrated embodiment, an application or website related to a specified service may be running on the computer device 100b, and the computer device 100a is a server of the application or website. A user performs operations related to the specified service in the application or website, thereby generating time-series service data of the specified service.

[0032] In one illustrated embodiment, the computer device 100a can record the time-series business data of a specified business. After the computer device 100a obtains the time-series business data at the current moment, it can perform anomaly detection on the time-series business data at the current moment based on the time-series business data of the specified business history to determine whether an anomaly occurs at the current moment.

[0033] In an illustrated embodiment, the computer device 100b may record the time series business data of the specified business, the computer device 100a may obtain the time series business data from at least one computer device 100b, and perform anomaly detection on the time series business data at the current moment based on the time series business data of the specified business history to determine whether an anomaly occurs at the current moment. Alternatively, the computer device 100b actively uploads new time series business data to the computer device 100a, and the computer device 100a may perform anomaly detection on the time series business data at the current moment based on the time series business data of the specified business history to determine whether an anomaly occurs at the current moment.

[0034] In an illustrated embodiment, if the computer device 100a detects that an abnormality has occurred at the current moment, the computer device 100a may also backtrack a period of time before the current moment to analyze the cause of the abnormality, that is, perform a root cause analysis.

[0035] Exemplarily, after the computer device 100a detects an abnormality at the current moment, an abnormality detection is performed every minute in a time period of 30 to 60 minutes before the current moment, and a root cause analysis is performed based on the abnormality detection results of every minute in the time period.

[0036] In an illustrated embodiment, when the computer device 100a traces back a period of time before the current moment, it can determine the first time period and the second time period in the historical reference time period of the time period to be analyzed based on the time period to be analyzed of the specified service; wherein the historical reference time period includes multiple cycles, the first time period includes the same phase time period of the time period to be analyzed in each cycle, and the second time period is the other time period in the historical reference time period except the first time period. The computer device 100a obtains the time series service data of the specified service in the first time period according to the first acquisition method to obtain the first time series service data, and obtains the time series service data of the specified service in the second time period according to the second acquisition method to obtain the second time series service data; wherein the data acquisition interval of the first acquisition method is less than the data acquisition interval of the second acquisition method. The computer device 100a performs data analysis based on the acquired time series service data to obtain the analysis result corresponding to the time period to be analyzed.

[0037] As described above, this manual does not need to use the abnormal point and the data points of each minute in a period of time before the abnormal point as detection points in sequence, and repeatedly perform multiple abnormality detections. Instead, only one data analysis is required to obtain the abnormality detection results of multiple detection points in a period of time before the abnormal point, thereby realizing batch abnormality detection at a faster detection speed.

[0038] In one illustrated embodiment, the computer device 100b may be a smart wearable device, a smart phone, a tablet computer, a laptop computer, a desktop computer, a vehicle-mounted computer, etc. having the above functions, and this specification does not make any specific limitation on this.

[0039] In one illustrated embodiment, the computer device 100a may be a server having the above functions, or may be a server cluster or a cloud computing service center composed of multiple servers, etc., and this specification does not make any specific limitation on this.

[0040] The data analysis method provided in this specification can be applied to multiple scenarios such as root cause analysis and data prediction. This specification does not limit the application scenarios of this specification, and only uses the following two examples for illustrative purposes:

[0041] For example, the data analysis method provided in this manual can be applied to the root cause analysis scenario:

[0042] Root cause analysis is a systematic problem-solving process that aims to identify and analyze the cause of the problem, find solutions to the problem, and specify preventive measures for the problem. When an abnormality is detected in a specified business at the current moment, an abnormality detection can be performed for each minute in the time period of 30 to 60 minutes before the current moment based on the time series business data of the specified business within 5 days before the current moment, and a root cause analysis can be performed based on the abnormality detection results of each minute in this period. If the data analysis method provided in this specification is adopted, the 30 to 60 minutes before the current moment can be used as the time period to be analyzed, and the time period of 5 days before the current moment can be used as the historical reference time period; after executing the data analysis method provided in this specification, the analysis results of multiple minutes in the 30 to 60 minutes before the current moment can be obtained. Therefore, through the data analysis method provided in this specification, the abnormality detection results of multiple moments in the time period to be analyzed can be obtained at one time, and the root cause analysis can be performed based on the abnormality detection results of the multiple moments, thereby improving the efficiency of the root cause analysis.

[0043] For example, the data analysis method provided in this specification can be applied to data prediction scenarios:

[0044] Data prediction is a method of modeling historical time series data to predict future time points. Exemplarily, hotel reservation data for the next X (any integer greater than 0) months can be predicted based on the hotel reservation data for the previous N (any integer greater than 1) years. If the data analysis method provided in this specification is used, the time period of the previous N years can be used as a historical reference time period, and the next X months can be used as a time period to be analyzed; then the data analysis method provided in this specification is executed to obtain hotel reservation data for multiple moments in the next X months, that is, hotel reservation data for multiple months in the next X months can be predicted at one time, or hotel reservation data for multiple days in the next month can be predicted at one time, thereby improving prediction efficiency.

[0045] See also Figure 2 , Figure 2 is a flowchart of a data analysis method provided by an exemplary embodiment. The method can be applied to a computer device or Figure 1 In the system architecture shown in Figure 2 As shown, the method may specifically include the following steps S201 to S203.

[0046] Step S201, based on the time period to be analyzed of the specified business scenario, determine the first time period and the second time period in the historical reference time period of the time period to be analyzed, the historical reference time period includes multiple cycles, the first time period includes the same phase time period of the time period to be analyzed in each cycle, and the second time period is the other time period in the historical reference time period except the first time period.

[0047] In an illustrated implementation, the designated service may be any service provided by an application program, for example, a payment service, a chat service, a financial service, a travel service, a hotel service, etc., which is not specifically limited in this specification.

[0048] In an illustrated implementation, the time period to be analyzed of the specified service is used to indicate that the relevant situation of the specified service in the time period needs to be analyzed. The time period to be analyzed may be a time period that has already occurred, that is, a historical time period; of course, the time period to be analyzed may also be a time period that has not yet occurred, that is, a future time period, and this specification does not specifically limit this.

[0049] Exemplarily, when the time period to be analyzed is a historical time period, the data analysis of the specified business in this specification may be to analyze whether there is an abnormality in the specified business in the time period to be analyzed. Exemplarily, when the time period to be analyzed is a future time period, the data analysis of the specified business in this specification may be to predict the situation of the specified business in the time period to be analyzed.

[0050] In an illustrated embodiment, the time period to be analyzed is determined by a computer device according to preset rules. Exemplarily, when the computer device determines that there is an abnormality at the current moment, it determines the current moment as the end time point of the time period to be analyzed; based on the end time point of the time period to be analyzed and the preset time length of the time period to be analyzed, the start time point of the time period to be analyzed is determined. Exemplarily, the computer device obtains a sequence of time periods to be analyzed, and the sequence of time periods to be analyzed includes multiple time periods to be analyzed; when any time period to be analyzed meets the analysis condition, the computer device executes the data analysis method provided in this specification based on the time period to be analyzed that meets the analysis condition. Among them, the analysis condition can be that the current moment is the start time point of the time period to be analyzed, or it can also be that the current moment is the end time point of the time period to be analyzed.

[0051] In another illustrated embodiment, the time period to be analyzed is set by the user. Exemplarily, the computer device displays a setting interface for the time period to be analyzed; the user can enter the time period to be analyzed in the setting interface; and the computer device obtains the input time period to be analyzed based on the setting interface.

[0052] In an illustrated embodiment, the historical reference time period is the time period to which the time series business data to be referenced belongs when analyzing the time series business data of the time period to be analyzed. The historical reference time period includes the time period before the time period to be analyzed. Moreover, the historical reference time period may include the time period to be analyzed or may not include the time period to be analyzed, which is not limited in this specification.

[0053] Exemplarily, when the time period to be analyzed is a historical time period, the historical reference time period may include the time period to be analyzed. That is, the end time point of the historical reference time period is the end time point of the time period to be analyzed. Exemplarily, when the time period to be analyzed is a future time period, the historical reference time period may not include the time period to be analyzed. That is, the end time point of the historical reference time period is the start time point of the time period to be analyzed.

[0054] In an illustrated embodiment, the multiple periods included in the historical reference time period may be time periods. That is, the historical reference time period includes multiple time periods. The time periods may be years, months, days, hours, minutes, etc.

[0055] Through experiments, it is found that when the time series business data has the following conditions, executing the data analysis method provided in this specification according to the time period can obtain better data analysis results: when the time series business data of a specified business at any moment includes multiple indicator data, some indicator data are periodic, and some indicator data are not periodic; or, multiple indicator data are not periodic.

[0056] In another illustrated embodiment, when the time series business data of the specified business is periodic data, the multiple periods included in the historical reference time period may be data change periods of the time series business data. It should be noted that this specification is only an illustrative description of the multiple periods included in the historical reference time period, and does not limit the conditions for using the periods. When the time series business data of the specified business is periodic, a time period may also be used, and this specification does not limit this.

[0057] In an illustrated embodiment, the historical reference time period includes multiple time periods. Based on the time period to be analyzed of the specified business scenario, determining a first time period in the historical reference time period of the time period to be analyzed includes: determining the same phase time period of the time period to be analyzed in each time period in the historical reference time period as the first time period.

[0058] Exemplarily, taking the time period of days as an example, when the time period to be analyzed is from 2:00 p.m. to 3:00 p.m. on November 19 and the historical reference time period is from 3:00 p.m. on November 14 to 3:00 p.m. on November 19, the time period from 2:00 p.m. to 3:00 p.m. on each day in the historical reference time period is determined as the first time period.

[0059] In another illustrated embodiment, the historical reference time period includes multiple data change cycles. The computer device can perform periodic detection on the time series service data of the specified service to obtain the period information of the time series service data. The period information may include information such as the period length, the period start point, and the period end point. This specification does not limit the period information.

[0060] Exemplarily, the number of data change cycles included in the historical reference time period is preset. The computer device can determine the historical reference time period of the time period to be analyzed based on the time period to be analyzed and the cycle information, so that the historical reference time period includes a preset number of data change cycles.

[0061] Exemplarily, the length of the historical reference time period is preset. The computer device can determine the historical reference time period based on the time period to be analyzed and the length of the historical reference time period preset. For example, the starting time point or the ending time point of the time period to be analyzed is determined as the ending time point of the historical reference time period; based on the ending time point of the historical reference time period and the length of the historical reference time period preset, the starting time point of the historical reference time period is determined.

[0062] Exemplarily, the time length of the historical reference time period is determined based on the time length of the time period to be analyzed. The time length of the time period to be analyzed is proportional to the time length of the historical reference time period. For example, based on the time length of the time period to be analyzed, determining the time length of the historical reference time period includes: obtaining the time length of the time period to be analyzed from a preset corresponding relationship, and the preset corresponding relationship includes a corresponding relationship between the time lengths of multiple time periods to be analyzed and the time lengths of the historical reference time periods. For another example, based on the time length of the time period to be analyzed, determining the time length of the historical reference time period includes: determining the time length of the historical reference time period based on the time length of the time period to be analyzed and a preset relationship function, and the preset relationship function is used to represent the relationship between the time length of the time period to be analyzed and the time length of the historical reference time period.

[0063] Exemplarily, the historical reference time period is set by the user. For example, the computer device displays a setting interface for the historical reference time period; the user can input the historical reference time period in the setting interface; and the computer device obtains the input historical reference time period based on the setting interface.

[0064] After determining the historical reference time period, the computer device may determine the same phase time period of the time period to be analyzed in each data change cycle in the historical reference time period as the first time period based on the period information of the time series business data of the historical reference time period.

[0065] Step S202: Acquire the time-series business data of the specified business in the first time period according to the first acquisition method to obtain the first time-series business data; and acquire the time-series business data of the specified business in the second time period according to the second acquisition method to obtain the second time-series business data. The data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method.

[0066] In an illustrated implementation manner, the business data of the designated business is time-series, and therefore, the business data of the designated business is referred to as time-series business data in this description.

[0067] In an illustrated embodiment, the time series business data of a designated business at any time includes multiple indicator data. The multiple indicator data may include at least one facility indicator data and at least one business indicator data. The at least one facility indicator data may include CPU memory usage, network bandwidth, network throughput, etc. Taking the designated business as a payment business as an example, the business indicator data may include order quantity, order amount, transaction amount, etc.

[0068] In one illustrated embodiment, for each indicator data in the time series business data, the above steps S202 to S203 are respectively executed to obtain analysis results corresponding to each indicator data. In another illustrated embodiment, the time series business data obtained in the above step S202 includes multiple indicator data, and in the following step S203, the multiple indicator data are analyzed together to obtain analysis results corresponding to the multiple indicator data.

[0069] In an illustrated embodiment, the first acquisition method is full acquisition, and the second acquisition method is sampling acquisition. Therefore, the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method. Among them, acquiring the time series business data of the specified business in the first time period according to the first acquisition method to obtain the first time series business data includes: acquiring the time series business data of the specified business in the first time period in full to obtain the first time series business data. Acquiring the time series business data of the specified business in the second time period according to the second acquisition method to obtain the second time series business data includes: acquiring the time series business data of the specified business in the second time period by sampling according to the second data collection interval to obtain the second time series business data. By fully acquiring the time series business data of the same phase time period, the data analysis results of each moment in the time period to be analyzed can be obtained when performing data analysis later.

[0070] The second data collection interval may be any non-zero interval, for example, 10, 15, 20, etc.

[0071] In an illustrated embodiment, the first acquisition method is sampling acquisition, and the second acquisition method is also sampling acquisition, but the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method. Acquiring the time series service data of the specified service in the first time period according to the first acquisition method to obtain the first time series service data includes: sampling and acquiring the time series service data of the specified service in the first time period according to the first data collection interval to obtain the first time series service data. Acquiring the time series service data of the specified service in the second time period according to the second acquisition method to obtain the second time series service data includes: sampling and acquiring the time series service data of the specified service in the second time period according to the second data collection interval to obtain the second time series service data.

[0072] The first data collection interval is smaller than the second data collection interval. For example, the first data collection interval is 2, and the second data collection interval is 10. Since the data collection interval used in the same phase time period is smaller, when performing data analysis later, analysis results of more moments in the time period to be analyzed can be obtained.

[0073] In one illustrated implementation, the computer device may obtain the time series service data from the log data. In another illustrated implementation, the computer device may obtain the time series service data from a database of a specified service. This specification does not limit the data source of the time series service data.

[0074] Step S203: performing data analysis based on the acquired time series business data to obtain analysis results corresponding to the time period to be analyzed.

[0075] In an illustrated embodiment, data analysis may be anomaly detection. Performing data analysis based on the acquired time series business data to obtain an analysis result corresponding to the time period to be analyzed includes: when the time period to be analyzed is a historical time period, performing anomaly detection on the time series business data of the time period to be analyzed based on the acquired time series business data to obtain anomaly detection results corresponding to the time period to be analyzed.

[0076] In an illustrated embodiment, data analysis may be data prediction. Data analysis is performed based on the acquired time series business data to obtain an analysis result corresponding to the time period to be analyzed, including: when the time period to be analyzed is a future time period, based on the acquired time series business data, predicting the time series business data of the time period to be analyzed, and obtaining a prediction result corresponding to the time period to be analyzed.

[0077] In one illustrated embodiment, the computer device may use any of the following data analysis methods when performing data analysis: 3sigma (a method for measuring the discreteness of data and identifying outliers); isolationtree (isolation tree, a binary tree structure used to isolate each observation); xgboost (eXtreme GradientBoosting, an algorithm based on gradient boosting decision trees); RNN (Recurrent Neural Network).

[0078] It should be noted that although this specification has improved the data to be processed for data analysis, the improved solution is still applicable to the current data analysis algorithm. In other words, the data analysis method provided in this specification can be implemented without improving the data analysis algorithm.

[0079] In summary, this specification obtains the time series business data of the first time period according to the first acquisition method, and obtains the time series business data of the second time period according to the second acquisition method, so as to obtain the time series business data used for data analysis. Since the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method, and the first time period is the same phase time period as the time period to be analyzed, more time series business data closely related to the time period to be analyzed is retained, so it can be used to perform batch analysis on multiple moments within the time period to be analyzed, and ensure the accuracy of the batch analysis, without the need to separately and repeatedly obtain and analyze the time series business data for multiple moments, thereby improving the overall data analysis efficiency. Therefore, by applying the data analysis method provided in this specification to anomaly detection scenarios, batch anomaly detection can be achieved at a faster detection speed.

[0080] See also Figure 3 , Figure 3 is a flowchart of a data analysis method provided by an exemplary embodiment. The method can be applied to a computer device or Figure 1 In the system architecture shown in Figure 3 As shown, the method may specifically include the following steps S301 to S303.

[0081] Step S301, based on the time period to be analyzed of the specified service, determine the first time period and the second time period in the historical reference time period of the time period to be analyzed, the historical reference time period includes multiple cycles, the first time period includes adjacent time periods in each cycle, the adjacent time periods are adjacent to the starting time point of the same phase time period in the same cycle, the same phase time period is the time period in each cycle that is in the same phase as the time period to be analyzed, and the second time period is the other time period in the historical reference time period except the first time period.

[0082] It should be noted that the above step 301 is similar to the above step 201, and can refer to the above step S201, which will not be described one by one here, and only the differences between step S301 and step S201 will be described.

[0083] The first time period determined in the above step 201 is an in-phase time period, and the first time period determined in step S301 is an adjacent time period. Next, the process of determining the adjacent time periods is exemplarily described.

[0084] In an illustrated embodiment, the time length of the adjacent time period is determined based on the time length of the time period to be analyzed, and the time length of the adjacent time period is positively correlated with the time length of the time period to be analyzed. Wherein, based on the time period to be analyzed of the specified business scenario, the adjacent time periods in each cycle are determined, including: based on the time length of the time period to be analyzed, determining the neighborhood time length, the neighborhood time length is used to indicate the time length of the adjacent time period to be determined; based on the starting time point of the same phase time period in each cycle and the neighborhood time length, determining the adjacent time period in each cycle.

[0085] Exemplarily, the time length of the time period to be analyzed is 60 minutes, and the corresponding neighborhood time length is 40 minutes. For any cycle, the starting time point of the same-phase time period in the cycle is used as the ending time point of the adjacent time period in the cycle, and the starting time point of the same-phase time period is moved forward 40 minutes to obtain the starting time point of the adjacent time period.

[0086] In an illustrated embodiment, a computer device is preset with a correspondence relationship between the time length of a plurality of time periods to be analyzed and the neighborhood time length, and the computer device determines the neighborhood time length based on the correspondence relationship. Wherein, determining the neighborhood time length based on the time length of the time period to be analyzed includes: obtaining the neighborhood time length corresponding to the time length of the time period to be analyzed from the preset correspondence relationship, and the preset correspondence relationship includes the correspondence relationship between the time length of a plurality of time periods to be analyzed and the neighborhood time length.

[0087] For example, the preset corresponding relationship may be determined based on experiments, may be set by a user, or may be an empirical value, which is not limited in this specification. For example, the time length of the time period to be analyzed is 60 minutes, and the corresponding neighborhood time length is 40 minutes.

[0088] In an illustrated embodiment, there is a functional relationship between the time length of the time period to be analyzed and the neighborhood time length. Determining the neighborhood time length based on the time length of the time period to be analyzed includes: determining the neighborhood time length based on the time length of the time period to be analyzed and a time length relationship function, wherein the time length relationship function is used to represent the relationship between the time length of the time period to be analyzed and the neighborhood time length.

[0089] Exemplarily, multiple time length combinations are determined through experiments, and one time length combination includes a time length of a time period to be analyzed and a neighborhood time length; the neighborhood time length included in the time length combination is the neighborhood time length obtained based on the accuracy of the analysis results during the experiment under the corresponding time period to be analyzed; by fitting the multiple time length combinations, the above-mentioned time length relationship function can be obtained. Exemplarily, the time length relationship function is set by the user. Exemplarily, the time length relationship function is an empirical function. This specification does not limit the time length relationship function.

[0090] It should be noted that this specification only takes the historical reference time period including multiple cycles as an example to exemplify the first time period, and in another illustrated embodiment, the historical reference time period may be independent of the cycle, that is, there is no need to limit the historical reference time period to include multiple cycles. The first time period is a time period adjacent to the starting time point of the time period to be analyzed.

[0091] Another point that needs to be explained is that this specification only exemplifies the data analysis method by analyzing the situation of a specified business in a certain time period. In another illustrated embodiment, the situation of a specified business at a certain point in time can also be analyzed. Accordingly, based on the time point to be analyzed of the specified business, a first time period and a second time period in the historical reference time period of the time point to be analyzed are determined, wherein the first time period includes a time period that is before the time point to be analyzed and adjacent to the time point to be analyzed, and the second time period is a time period other than the first time period.

[0092] Exemplarily, based on the time point to be analyzed of the specified service, when determining the first time period in the historical reference time period of the time point to be analyzed, the time length of the first time period is a fixed value. The time length of the first time period can be determined based on experiments, can be set by the user, or can be an empirical value. This specification does not limit the time length of the first time period.

[0093] Step S302: Acquire the time-series business data of the specified business scenario in the first time period according to the first acquisition method to obtain the first time-series business data; and acquire the time-series business data of the specified business scenario in the second time period according to the second acquisition method to obtain the second time-series business data. The data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method.

[0094] The above step 302 is similar to the above step 202, and can refer to the above step S202, which will not be described in detail here.

[0095] Step S303: using a data analysis algorithm to perform data analysis based on the acquired time series business data, and obtaining analysis results corresponding to the time period to be analyzed.

[0096] The above step 303 is the same as the above step 203, and can refer to the above step S203, which will not be described in detail here.

[0097] In summary, this specification obtains the time series business data of the first time period according to the first acquisition method, and obtains the time series business data of the second time period according to the second acquisition method, so as to obtain the time series business data used for data analysis. Since the first time period is an adjacent time period of the time period to be analyzed, the prediction features of the time period to be analyzed can be more accurately constructed based on the time series business data of the first time period. Since the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method, more time series business data that can accurately construct the prediction features of the time period to be analyzed are retained, so the accuracy of the analysis results can be improved.

[0098] See also Figure 4 , Figure 4 is a flowchart of a data analysis method provided by an exemplary embodiment. The method can be applied to a computer device or Figure 1 In the system architecture shown in Figure 4 As shown, the method may specifically include the following steps S401 to S403.

[0099] Step S401, based on the time period to be analyzed of the specified service, determine the first time period and the second time period in the historical reference time period of the time period to be analyzed, the historical reference time period includes multiple cycles, the first time period includes the same phase time period of the time period to be analyzed in each cycle, the first time period also includes adjacent time periods in each cycle, the adjacent time periods are adjacent to the starting time point of the same phase time period in the same cycle, and the second time period is the other time period in the historical reference time period except the first time period.

[0100] It should be noted that the above step S401 is similar to the above step S201 and step S301, and reference may be made to the above step S201 and step S301, which will not be described one by one here.

[0101] Step S402: Acquire the time-series business data of the specified business scenario in the first time period according to the first acquisition method to obtain the first time-series business data; and acquire the time-series business data of the specified business scenario in the second time period according to the second acquisition method to obtain the second time-series business data. The data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method.

[0102] It should be noted that the above step S402 is similar to the above step S202, and can refer to the above step S202, which will not be described one by one here. Since the first time period in the above step S402 includes the same phase time period and the adjacent time period, and the first time period in the above step S202 includes the same phase time period, therefore, the above step S402 is supplemented based on this difference:

[0103] When the computer device acquires the time series service data of the same phase time period and the time series service data of the adjacent time period, the data acquisition intervals adopted may be the same or different. In an illustrated embodiment, the computer device acquires the time series service data of the specified service in the first time period according to the first acquisition method to obtain the first time series service data, including: acquiring the time series service data of the same phase time period of each cycle and the time series service data of the adjacent time period of each cycle according to the same data acquisition interval to obtain the first time series service data.

[0104] Exemplarily, the timing service data of the same phase time period and the adjacent time period can be fully acquired. According to the same data collection interval, the timing service data of the specified service data in the same phase time period of each cycle and the timing service data of the adjacent time period of each cycle are acquired to obtain the first timing service data, including: fully acquiring the timing service data of the specified service in the same phase time period of each cycle and the timing service data of the adjacent time period of each cycle to obtain the first timing service data. By fully acquiring the timing service data of the same phase time period and the adjacent time period, not only can the analysis results of each moment in the time period to be analyzed be analyzed, but also the accuracy of the analysis results can be improved as much as possible.

[0105] Exemplarily, the data analysis method of this specification is compared with the single-point anomaly detection method and the full-data anomaly detection method in the prior art through the following examples:

[0106] In the intelligent operation and maintenance scenario, after obtaining the time series business data of the specified business at the current moment, the time series business data of the specified business in the previous 5 days is used to detect whether there are any anomalies in the time series business data at the current moment. Since the time series business data of the specified business is minute-level data, if full anomaly detection is performed, the data length for each anomaly detection is 7200 (there are 1440 minutes in a day, so there are 1440 data points in a day, and 1440*5=7200 data points in 5 days).

[0107] If a single-point anomaly detection method is used, a point can be selected every 10 points from the current time x(t) in the historical direction, such as Figure 5As shown in Figure (1), after obtaining 5 days of time series business data, anomaly detection is performed on the current time x(t). At this time, the data length of single-point anomaly detection is 720=7200 / 10.

[0108] If an anomaly is detected at the current time x(t), the anomaly detection of 60 minutes before x(t) can be traced back. Since the time series business data of 60 minutes before x(t) is "lost" in the sampling, if you want to trace the anomaly detection of 60 minutes before x(t), you need to use each minute of the 60 minutes before x(t) as a detection point and perform 60 heavy anomaly detections.

[0109] At this time, if the data analysis method provided in this manual is used, the adjacent time periods are taken as 40 minutes, and the full amount is acquired for the same phase time period and the adjacent time period, and the sampling acquisition is still carried out in the manner of 1:10 for other time periods, such as Figure 5 As shown in Figure (2), the data length of anomaly detection is: (40+60+(1440-(40+60)) / 10)*5=1170. Compared with the single-point anomaly detection method, the data length is increased, but only one anomaly detection is required to obtain the anomaly detection results of each minute in the first 60 minutes of x(t), without repeated execution. Therefore, the overall anomaly detection speed is greatly improved.

[0110] In addition, if full anomaly detection is performed, the anomaly detection results for each minute in the first 60 minutes of x(t) can also be obtained at one time, but the data length of full anomaly detection is 7200, while the data length of the data analysis method provided in this specification is 1170. Obviously, the data analysis method provided in this specification has a faster anomaly detection speed. Moreover, compared with full anomaly detection, the anomaly deceleration speed of the data analysis method provided in this specification is increased by about 10 times.

[0111] It should be noted that Figure 5 The X-axis is the time period, the Y-axis is the time series business data axis, and, Figure 5 "Circles" are used to represent the time series service data of the second time period acquired by sampling, "squares" are used to represent the time series service data of the same phase time period acquired by the whole quantity, and "triangles" are used to represent the time series service data of the adjacent time period acquired by the whole quantity. Figure 5 "Circles", "squares" and "triangles" are only used to distinguish the same-phase time period, adjacent time period and second time period. The number of "circles" does not really represent the number of time series service data acquired by sampling. Similarly, the number of "squares" and "triangles" does not really represent the number of time series service data acquired in full. Figure 5 It is only used to clarify the relationship between the same phase time period and adjacent time periods in each cycle. Figure 5The period in does not really represent the period of the time series business data. Figure 5 The data length in does not truly represent the data length of the acquired time series business data.

[0112] Exemplarily, the timing service data of the same phase time period and the adjacent time period can be sampled and obtained. According to the same data collection interval, the timing service data of the same phase time period of each cycle of the specified service and the timing service data of the adjacent time period of each cycle are obtained to obtain the first timing service data, including: according to the first data collection interval, the timing service data of the same phase time period of each cycle of the specified service and the timing service data of the adjacent time period of each cycle are sampled and obtained to obtain the first timing service data. By sampling and obtaining the timing service data of the same phase time period and the adjacent time period, the data length can be reduced as much as possible and the data analysis speed can be improved.

[0113] In another illustrated embodiment, a computer device acquires timing service data of a specified service in a first time period according to a first acquisition method to obtain first timing service data, including: according to two different data collection intervals, respectively acquiring the timing service data of the specified service in the same phase time period of each cycle and the timing service data in adjacent time periods of each cycle to obtain first timing service data.

[0114] Exemplarily, the timing service data of the same phase time period can be fully acquired, and the timing service data of the adjacent time period can be sampled. In this way, the analysis results corresponding to each moment in the time period to be analyzed can be obtained, so as to better perform the root cause analysis later, and the accuracy of the analysis results can also be improved. For example, according to two different data collection intervals, the timing service data of the same phase time period of each cycle of the specified service and the timing service data of the adjacent time period of each cycle are respectively acquired to obtain the first timing service data, including: the timing service data of the same phase time period of each cycle of the specified service is fully acquired to obtain the first sub-timing service data in the first timing service data, and the timing service data of the adjacent time period of each cycle of the specified service is sampled and acquired according to the third data collection interval to obtain the second sub-timing service data in the first timing service data. Among them, the third data collection interval is less than the second data collection interval. In addition, the third data collection interval and the first data collection interval can be the same or different.

[0115] Exemplarily, the timing service data of the same phase time period can be sampled and the timing service data of the adjacent time period can be fully acquired. That is, more emphasis is placed on improving the accuracy of the analysis results. For example, according to two different data collection intervals, the timing service data of the same phase time period of each cycle and the timing service data of the adjacent time period of each cycle of the specified service are respectively acquired to obtain the first timing service data, including: sampling and acquiring the timing service data of the same phase time period of each cycle of the specified service according to the fourth data collection interval, obtaining the first sub-timing service data in the first timing service data, and fully acquiring the timing service data of the adjacent time period of each cycle of the specified service, obtaining the second sub-timing service data in the first timing service data. Among them, the fourth data collection interval is smaller than the second data collection interval. In addition, the fourth data collection interval and the first data collection interval can be the same or different.

[0116] Exemplarily, the time series service data of the same phase time period and the time series service data of the adjacent time period can be obtained according to two different data collection intervals. For example, according to two different data collection intervals, the time series service data of the same phase time period of each cycle and the time series service data of the adjacent time period of each cycle of the specified service are respectively obtained to obtain the first time series service data, including: sampling and obtaining the time series service data of the same phase time period of each cycle of the specified service according to the fifth data collection interval, obtaining the first sub-time series service data in the first time series service data, and sampling and obtaining the time series service data of the adjacent time period of each cycle of the specified service according to the sixth data collection interval, obtaining the second sub-time series service data in the first time series service data. Among them, the fifth data collection interval and the sixth data collection interval are both smaller than the second data collection interval. Moreover, the fifth data collection interval and the sixth data collection interval are different, and the sizes of the fifth data collection interval and the sixth data collection interval can be determined according to actual needs. If more emphasis is placed on accuracy, the sixth data collection interval is smaller than the fifth data collection interval; if more emphasis is placed on obtaining analysis results at more moments, the fifth data collection interval is smaller than the sixth data collection interval.

[0117] Step S403: using a data analysis algorithm, performing data analysis based on the acquired time series business data, and obtaining analysis results corresponding to the time period to be analyzed.

[0118] The above step 403 is similar to the above step 203, and can refer to the above step S203, which will not be described in detail here.

[0119] In summary, this specification obtains the time series business data of the first time period according to the first acquisition method, and obtains the time series business data of the second time period according to the second acquisition method, so as to obtain the time series business data used for data analysis. Since the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method, and the first time period includes the same phase time period as the time period to be analyzed, more time series business data closely related to the time period to be analyzed are retained, so it can be used to perform batch analysis on multiple moments within the time period to be analyzed, and ensure the accuracy of batch analysis, without the need to separately and repeatedly obtain and analyze the time series business data for multiple moments, thereby improving the overall data analysis efficiency. Therefore, by applying the data analysis method provided in this specification to anomaly detection scenarios, batch anomaly detection can be achieved at a faster detection speed.

[0120] Furthermore, since the first time period also includes the adjacent time periods of the time period to be analyzed, the time series business data based on the first time period can more accurately construct the prediction features of the time period to be analyzed. Since the data collection interval of the first acquisition method is smaller than the data collection interval of the second acquisition method, more time series business data that can accurately construct the prediction features of the time period to be analyzed are retained, thereby improving the accuracy of the analysis results.

[0121] Figure 6 is a schematic structural diagram of a device provided by an exemplary embodiment. Figure 6 At the hardware level, the device includes a processor 602, an internal bus 604, a network interface 606, a memory 608, and a non-volatile memory 610, and may also include hardware required for other functions. One or more embodiments of this specification may be implemented based on software, such as the processor 602 reading the corresponding computer program from the non-volatile memory 610 into the memory 608 and then running it. Of course, in addition to the software implementation, one or more embodiments of this specification do not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0122] Please refer to Figure 7 , the data analysis device can be applied to Figure 6 The device shown in the figure is used to implement the technical solution of this specification. The data analysis device may include:

[0123] The time period determining unit 701 is configured to determine, based on the time period to be analyzed of the specified service, a first time period and a second time period in a historical reference time period of the time period to be analyzed, wherein the historical reference time period includes a plurality of cycles, the first time period includes a same-phase time period of the time period to be analyzed in each cycle, and the second time period is a time period other than the first time period in the historical reference time period;

[0124] The data acquisition unit 702 is used to acquire the time series service data of the specified service in the first time period according to a first acquisition method to obtain the first time series service data, and to acquire the time series service data of the specified service in the second time period according to a second acquisition method to obtain the second time series service data, wherein the data acquisition interval of the first acquisition method is smaller than the data acquisition interval of the second acquisition method;

[0125] The data analysis unit 703 is used to perform data analysis based on the acquired time series business data to obtain an analysis result corresponding to the time period to be analyzed.

[0126] In an illustrated embodiment, the data acquisition unit 702 is specifically configured to:

[0127] Acquire all the time series service data of the designated service in the first time period to obtain the first time series service data; or, acquire the time series service data of the designated service in the first time period by sampling according to the first data collection interval to obtain the first time series service data;

[0128] According to the second data collection interval, sample and obtain the time series service data of the designated service in the second time period to obtain the second time series service data.

[0129] In an illustrated embodiment, the first time period also includes adjacent time periods in each cycle, and the adjacent time periods are adjacent to the starting time points of the same-phase time periods in the same cycle.

[0130] In an illustrated embodiment, the data acquisition unit 702 is specifically configured to:

[0131] Acquire the time series service data of the designated service in the same phase time period of each cycle and the time series service data of the designated service in the adjacent time period of each cycle according to the same data collection interval to obtain the first time series service data; or,

[0132] According to two different data collection intervals, the time series service data of the designated service in the same phase time period of each cycle and the time series service data in the adjacent time period of each cycle are respectively acquired to obtain the first time series service data.

[0133] In an illustrated embodiment, the data acquisition unit 702 is specifically configured to:

[0134] Acquire the time series service data of the designated service in the same phase time period of each cycle in full, and obtain the first sub-time series service data in the first time series service data; and acquire the time series service data of the designated service in the adjacent time period of each cycle by sampling according to the third data collection interval, and obtain the second sub-time series service data in the first time series service data; or

[0135] Sampling the time series service data of the designated service in the same phase time period of each cycle according to the fourth data collection interval to obtain the first sub-time series service data in the first time series service data, and acquiring the time series service data of the designated service in the adjacent time period of each cycle in full to obtain the second sub-time series service data in the first time series service data; or,

[0136] The timing service data of the designated service in the same phase time period of each cycle is sampled according to the fifth data collection interval to obtain the first sub-timing service data in the first timing service data, and the timing service data of the designated service in the adjacent time period of each cycle is sampled according to the sixth data collection interval to obtain the second sub-timing service data in the first timing service data.

[0137] In an illustrated embodiment, the time period determination unit 701 is specifically configured to:

[0138] Determine a neighborhood time length based on the time length of the time period to be analyzed, where the neighborhood time length is used to indicate the time length of the adjacent time period to be determined;

[0139] The adjacent time periods in each cycle are determined based on the starting time point of the in-phase time periods in each cycle and the neighborhood time length.

[0140] In an illustrated embodiment, the time period determination unit 701 is specifically configured to:

[0141] Acquire a neighborhood time length corresponding to the time length of the time period to be analyzed from a preset corresponding relationship, wherein the preset corresponding relationship includes a corresponding relationship between the time lengths of multiple time periods to be analyzed and the neighborhood time lengths; or,

[0142] The neighborhood time length is determined based on the time length of the time period to be analyzed and a time length relationship function, where the time length relationship function is used to represent the relationship between the time length of the time period to be analyzed and the neighborhood time length.

[0143] In an illustrated implementation, the period is a time period, or the period is a data change period of time-series service data.

[0144] In an illustrated embodiment, the data analysis unit 703 is specifically configured to:

[0145] In the case where the time period to be analyzed is a historical time period, anomaly detection is performed on the time series business data of the time period to be analyzed based on the acquired time series business data to obtain an anomaly detection result corresponding to the time period to be analyzed; or

[0146] In the case that the time period to be analyzed is a future time period, the time series service data of the time period to be analyzed is predicted based on the acquired time series service data to obtain a prediction result corresponding to the time period to be analyzed.

[0147] Based on the same concept as the above method, this specification also provides a computer device, including: a processor; a memory for storing processor-executable instructions; wherein the processor implements the steps of the method described in any of the above embodiments by running the executable instructions.

[0148] Based on the same concept as the above method, this specification also provides a computer-readable storage medium on which computer instructions are stored. When the instructions are executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0149] Based on the same concept as the above method, this specification also provides a computer program product, including a computer program / instruction, which implements the steps of the method described in any of the above embodiments when executed by a processor.

Claims

1. A data analysis method, comprising: Based on the time period to be analyzed of the designated service, determine a first time period and a second time period in a historical reference time period of the time period to be analyzed, wherein the historical reference time period includes a plurality of cycles, the first time period includes a time period in phase with the time period to be analyzed in each cycle, and the second time period is a time period other than the first time period in the historical reference time period; Acquire the time series service data of the designated service in the first time period in a first acquisition manner to obtain first time series service data, and acquire the time series service data of the designated service in the second time period in a second acquisition manner to obtain second time series service data, wherein the data acquisition interval of the first acquisition manner is smaller than the data acquisition interval of the second acquisition manner; Data analysis is performed based on the acquired time series business data to obtain analysis results corresponding to the time period to be analyzed.

2. According to the method of claim 1, the step of acquiring the time series service data of the specified service in the first time period in a first acquisition manner to obtain the first time series service data comprises: Acquire all the time series service data of the specified service in the first time period to obtain the first time series service data; Alternatively, sampling and acquiring the time series service data of the specified service in the first time period according to the first data collection interval to obtain the first time series service data; The acquiring the time-series service data of the designated service in the second time period according to the second acquisition method to obtain the second time-series service data includes: According to the second data collection interval, sample and obtain the time series service data of the designated service in the second time period to obtain the second time series service data. 3 . The method according to claim 1 , wherein the first time period further comprises adjacent time periods in each of the cycles, and the adjacent time periods are adjacent to starting time points of the same-phase time periods in the same cycle.

4. The method according to claim 3, wherein acquiring the time series service data of the specified service in the first time period in a first acquisition manner to obtain the first time series service data comprises: Acquire the time series service data of the designated service in the same phase time period of each cycle and the time series service data of the designated service in the adjacent time period of each cycle according to the same data collection interval to obtain the first time series service data; or According to two different data collection intervals, the time series service data of the designated service in the same phase time period of each cycle and the time series service data in the adjacent time period of each cycle are respectively acquired to obtain the first time series service data.

5. The method according to claim 4, wherein the step of acquiring the time series service data of the designated service in the same phase time period of each cycle and the time series service data of the designated service in the adjacent time period of each cycle respectively according to two different data collection intervals to obtain the first time series service data comprises: Acquire all the time series service data of the designated service in the same phase time period of each cycle to obtain the first sub-time series service data in the first time series service data, and acquire the time series service data of the designated service in the adjacent time period of each cycle by sampling according to the third data collection interval to obtain the second sub-time series service data in the first time series service data; or Sampling and acquiring the time series service data of the designated service in the same phase time period of each cycle according to the fourth data collection interval, obtaining the first sub-time series service data in the first time series service data, and acquiring the time series service data of the designated service in the adjacent time period of each cycle in full, obtaining the second sub-time series service data in the first time series service data; or, The timing service data of the designated service in the same phase time period of each cycle is obtained by sampling according to the fifth data collection interval, and the first sub-timing service data in the first timing service data is obtained; and the timing service data of the designated service in the adjacent time period of each cycle is obtained by sampling according to the sixth data collection interval, and the second sub-timing service data in the first timing service data is obtained.

6. The method according to claim 3, determining the adjacent time periods in each cycle based on the time period to be analyzed of the designated service, comprises: Determine a neighborhood time length based on the time length of the time period to be analyzed, where the neighborhood time length is used to indicate the time length of the adjacent time period to be determined; The adjacent time periods in each of the cycles are determined based on the starting time points of the in-phase time periods in each of the cycles and the neighborhood time lengths.

7. The method according to claim 6, wherein determining the neighborhood time length based on the time length of the time period to be analyzed comprises: Acquire a neighborhood time length corresponding to the time length of the time period to be analyzed from a preset corresponding relationship, wherein the preset corresponding relationship includes a corresponding relationship between the time lengths of multiple time periods to be analyzed and the neighborhood time lengths; or, The neighborhood time length is determined based on the time length of the time period to be analyzed and a time length relationship function, where the time length relationship function is used to represent the relationship between the time length of the time period to be analyzed and the neighborhood time length.

8. According to the method of claim 1, the period is a time period, or the period is a data change period of time series business data.

9. According to the method of claim 1, the step of performing data analysis based on the acquired time series business data to obtain analysis results corresponding to the time period to be analyzed comprises: In the case where the time period to be analyzed is a historical time period, based on the acquired time series business data, anomaly detection is performed on the time series business data of the time period to be analyzed to obtain an anomaly detection result corresponding to the time period to be analyzed; or, In the case that the time period to be analyzed is a future time period, the time series service data of the time period to be analyzed is predicted based on the acquired time series service data to obtain a prediction result corresponding to the time period to be analyzed.

10. A computer device comprising: Memory and processor; The memory stores a computer program executable by the processor; When the processor runs the computer program, the processor performs the method according to any one of claims 1 to 9.

11. A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method according to any one of claims 1 to 9.

12. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images