CN109145996A

CN109145996A - Achievement data generation method, device and electronic equipment under abnormal environment

Info

Publication number: CN109145996A
Application number: CN201811015492.6A
Authority: CN
Inventors: 吴蔚川; 朱光耀; 李弄潮
Original assignee: Alibaba Group Holding Ltd
Current assignee: Advanced New Technologies Co Ltd; Advantageous New Technologies Co Ltd
Priority date: 2018-08-31
Filing date: 2018-08-31
Publication date: 2019-01-04
Anticipated expiration: 2038-08-31
Also published as: CN109145996B

Abstract

The invention discloses achievement data generation method, device and the electronic equipment under a kind of abnormal environment, the method can the historical data to multiple indexs pre-process respectively, obtain the target data of the multiple index；Principal component analysis is carried out to the target data of the multiple index, obtains the data of multiple principal components；The distribution of data based at least one principal component and default confidence level determine the data of at least one principal component under abnormal environment, and default confidence level is the confidence level that the data of the multiple principal component belong to data under normal environment；To the data of at least one principal component described under the abnormal environment, the inverse operation and the pretreated inverse operation of the principal component analysis are successively carried out, prediction data of the multiple index under the abnormal environment is generated.

Description

Achievement data generation method, device and electronic equipment under abnormal environment

Technical field

This application involves the achievement data generation methods under field of computer technology more particularly to a kind of abnormal environment, dress It sets and electronic equipment.

Background technique

Bank's pressure test is a kind of method for assessing bank risk.Bank's pressure test includes that the credit pressure of bank is surveyed Examination, market pressure test, mobility pressure test and operating pressure test etc..In the pressure test of bank, need rationally may be used Multiple achievement datas under the abnormal environment of letter are as input parameter.For example, in the mobility pressure test of bank, due to upper Extra large interbank inter-bank offered rate (Shanghai Interbank Offered Rate, Shibor) can portray city well Mobility, therefore can be by a variety of term structures under abnormal environment (such as Shibor rise violently suddenly extreme market environment) Input parameter of the Shibor as bank liquidity pressure test.Wherein, the term structure of Shibor includes (O/N) overnight, one All (1W), two weeks (2W), 1 month (1M), 3 months (3M), 6 months (6M), 9 months (9M) and 1 year (1Y) 8 kinds.

But the extreme market environment to rise violently suddenly due to only occurring Shibor few in number in history, it needs Certain method prediction to be used generates the Shibor under more abnormal environments, as the defeated of bank liquidity pressure test Enter parameter.

Traditional method is that the data of multiple indexs under history extreme scenes are done with some amplitude adjustment, is generated more The data of multiple indexs under abnormal environment.This mode is too simple, does not account for the connection between different indexs.

Summary of the invention

The embodiment of the present application provides achievement data generation method, device and the electronic equipment under a kind of abnormal environment, with The connection between different achievement datas is taken into account, the reliability and reasonability of the multiple achievement datas predicted are improved.

In order to solve the above technical problems, the embodiment of the present application is achieved in that

In a first aspect, the achievement data generation method under proposing a kind of abnormal environment, which comprises

The historical data of multiple indexs is pre-processed respectively, obtains the target data of the multiple index, the mesh Mark the standardized data of the changing value between the historical data that data are different moments；

Principal component analysis is carried out to the target data of the multiple index, obtains the data of multiple principal components；

The distribution of data based at least one principal component and default confidence level determine described at least one under abnormal environment The data of a principal component, the default confidence level are the credible water that the data of the multiple principal component belong to data under normal environment It is flat；

To the data of at least one principal component described under the abnormal environment, the inverse of the principal component analysis is successively carried out Operation and the pretreated inverse operation, generate prediction data of the multiple index under the abnormal environment.

Second aspect, proposes the achievement data generating means under a kind of abnormal environment, and described device includes:

Preprocessing module pre-processes respectively for the historical data to multiple indexs, obtains the multiple index Target data, the target data are the standardized data of the changing value between the historical data of different moments；

Principal component analysis module carries out principal component analysis for the target data to the multiple index, obtains multiple masters The data of ingredient；

First determining module determines different for the distribution and default confidence level of the data based at least one principal component The data of at least one principal component under normal environment, the default confidence level is that the data of the multiple principal component belong to normally The confidence level of data under environment；

Generation module, for the data at least one principal component described under the abnormal environment, successively carry out described in The inverse operation of principal component analysis and the pretreated inverse operation, generate prediction of the multiple index under the abnormal environment Data.

The third aspect proposes a kind of electronic equipment, comprising:

Processor；And

It is arranged to the memory of storage computer executable instructions, the executable instruction makes the place when executed It manages device and executes following operation:

Fourth aspect proposes a kind of computer readable storage medium, the computer-readable recording medium storage one Or multiple programs, one or more of programs are when the electronic equipment for being included multiple application programs executes, so that the electricity Sub- equipment executes following operation:

As can be seen from the technical scheme provided by the above embodiments of the present application, scheme provided by the embodiments of the present application at least have as A kind of lower technical effect: since the target data to multiple indexs carries out the data for multiple principal components that principal component analysis obtains, Therefore the connection being able to reflect between the historical data of multiple indexs utilizes the index under principal component analysis predicted anomaly environment When data, the connection between different indexs can be taken into account, improves the reliability and reasonability of the multiple achievement datas predicted, When carrying out pressure test using the achievement data, obtained pressure testing results are also more accurate.

Detailed description of the invention

The drawings described herein are used to provide a further understanding of the present application, constitutes part of this application, this Shen Illustrative embodiments and their description please are not constituted an undue limitation on the present application for explaining the application.In the accompanying drawings:

Fig. 1 be this specification embodiment provide abnormal environment under achievement data generation method flow diagram it One.

Fig. 2 is the schematic illustration of the achievement data generation method under the abnormal environment that this specification embodiment provides.

Fig. 3 is multiple achievement data variation tendency schematic diagrames in Fig. 2 first quartile under extreme scenes.

Fig. 4 is multiple achievement data variation tendency schematic diagrames in Fig. 2 first quartile under extreme scenes.

Fig. 5 is multiple achievement data variation tendency schematic diagrames in Fig. 2 first quartile under extreme scenes.

Fig. 6 is multiple achievement data variation tendency schematic diagrames in Fig. 2 first quartile under extreme scenes.

Fig. 7 be this specification embodiment provide abnormal environment under achievement data generation method flow diagram it Two.

Fig. 8 is the normal distribution schematic diagram that this specification embodiment provides.

Fig. 9 is the structural schematic diagram for a kind of electronic equipment that this specification embodiment provides.

Figure 10 be this specification embodiment provide abnormal environment under achievement data generating means structural schematic diagram it One.

Figure 11 be this specification embodiment provide abnormal environment under achievement data generating means structural schematic diagram it Two.

Specific embodiment

To keep the purposes, technical schemes and advantages of the application clearer, below in conjunction with the application specific embodiment and Technical scheme is clearly and completely described in corresponding attached drawing.Obviously, described embodiment is only the application one Section Example, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art are not doing Every other embodiment obtained under the premise of creative work out, shall fall in the protection scope of this application.

Connection between different achievement datas in order to balance improves the reliability and rationally of the multiple achievement datas predicted Property, this specification provide achievement data generation method and device under a kind of abnormal environment.The abnormal environment that this specification provides Under achievement data generation method executing subject, can be server or terminal.

It should be noted that, although introducing this specification using bank's pressure test as application scenarios in the present specification Achievement data under the abnormal environment of offer generates scheme, it should be appreciated that the index number under the abnormal environment that this specification provides According to generation scheme can also be applied to other scenes.

It should also be noted that, in this specification embodiment normal environment and abnormal environment be in contrast, can be with Defined with the numerical value change trend of index, for example, it is assumed that Shibor interest rate overnight within a very long time be all 1%~ Fluctuate between 2% without there is big variation, illustrate at the Shibor interest rate overnight of this period of time in normal circumstances, On the basis of this, if Shibor interest rate overnight some day has risen violently 5%, the Shibor interest rate overnight of this day of specification is in Under abnormal environment (or extreme environment).

Achievement data generation side under a kind of abnormal environment that 1 to 8 pair of this specification embodiment provides with reference to the accompanying drawing Method is described in detail.

As shown in Figure 1, in one embodiment, the achievement data generation side under a kind of abnormal environment that this specification provides Method may include steps of:

Step 102 pre-processes the historical data of multiple indexs respectively, obtains the number of targets of the multiple index According to the target data is the standardized data of the changing value between the historical data of different moments.

In the present specification, index can be understood as to measure the standard of features level.For example, Bank of Shanghai Between inter-bank offered rate (Shanghai Interbank Offered Rate, Shibor) be one and can measure goods well The standard of the coin market liquidity.

It should also be understood that the index of description different things feature can be different, and the same feature of same thing is described Index can have multiple.For example, the index that can measure bank credit risk may include normal loan migration rate and bad loan Money migration rate two indices, this is different from the index of money market mobility is measured.For another example described in this specification background technique, The index S hibor that money market mobility can be measured can be with more than one, but including (O/N) overnight, one week (1W), two All (2W), 1 month (1M), 3 months (3M), 6 months (6M), the Shibor of 9 months (9M) and 1 year (1Y) 8 kinds of term structure.

In the present specification, for the convenience of description, mainly using the Shibor of structure different maturity periods 8 kinds as in step 102 Multiple indexs an example, carry out this specification offer each technical solution introduction.

In one example, step 102 can specifically include: to the historical data of the multiple index, respectively with default Time interval is that the posterior data of step size computation generation time (can also claim relative to the changing value of the preceding data of generation time For difference value), obtain the variation Value Data of the multiple index；Standard is carried out respectively to the variation Value Data of the multiple index Change processing, obtains the target data of the multiple index.Wherein, prefixed time interval can be set according to actual needs, For example, it is assumed that with the Shibor interest rate of following first day of prediction of historical data, when being used for bank's pressure test, preset time Interval can be set to one day；Assuming that with the Shibor interest rate of following second day of prediction of historical data, to be surveyed for bank's pressure When examination, prefixed time interval be can be set to two days, etc..

For example, it is assumed that the historical data of an index in multiple indexs is { a₁,a₂,a₃,a₄,…,a_i,…,a_n, In, a₁To a_nIt is according to the historical data acquired as unit of day, and a₁To a_nSequencing arrangement temporally, n are acquisition The quantity of historical data, i=1,2 ..., n.So, when prefixed time interval is 1 day, the changing value number of the obtained index According to for { a₂-a₁,a₃-a₂,a₄-a₃,…,a_i-a_i-1,…,a_n-a_n-1}。

Wherein, standardization may include normalized etc., can specifically be marked using Z score (zscore) Quasi-ization processing.It should be understood that standardization mode can there are many kinds of, be not listed one by one herein.

After being pre-processed using historical data of the above-mentioned pretreatment mode to multiple indexs, the mesh of obtained a certain index Mark data can be understood as, the standardized data of changing value of the index between the historical data of different moments.

Step 104 carries out principal component analysis to the target data of the multiple index, obtains the data of multiple principal components.

Principal component analysis (Principal Component Analysis, PCA) is a kind of method of mathematic(al) manipulation, its handle One group of given correlated variables changes into another group of incoherent variable by linear transformation, these new variables according to characteristic value according to The secondary sequence arrangement successively decreased, transformation, which obtains several variables just, several principal components.

At step 104, in addition to obtaining the data of multiple principal components, can also keep records of each principal component characteristic value and Transformation matrix etc., it is convenient to determine first principal component and Second principal component, etc. in subsequent steps, and facilitate subsequent led The inverse operation of constituent analysis.

It is to pass through line using multiple indexs as one group of correlated variables using principal component analysis in this specification embodiment Property transformation be converted into another group of incoherent variable, specific transform method can use the prior art, herein without detailed Description.For example, if multiple indexs are the Shibor of 8 kinds of term structures, their target data can be as shown in table 1, The data of the multiple principal components obtained by mathematic(al) manipulation can be as shown in table 2.It should be understood that in practical applications, the mesh in table 1 Marking data is some specific numerical value, and the data of the principal component in table 2 are also some specific numerical value.

Table 1

Serial number

It is overnight

One week

Two weeks

1 month

Three months

6 months

9 months

1 year

1

a₁

b₁

c₁

d₁

e₁

f₁

e₁

g₁

2

a₂

b₂

c₂

d₂

e₂

f₂

e₂

g₂

···

n

a_n

b_n

c_n

d_n

e_n

f_n

e_n

g_n

Table 2

Serial number	First principal component	Second principal component,	···	M principal component
					1	h₁	j₁	···	k₁
2	h₂	j₂	···	k₂
					···	···	···	···	···
n	h_n	j_n	···	k_n

The distribution and default confidence level of step 106, data based at least one principal component, determine institute under abnormal environment The data of at least one principal component are stated, the default confidence level is that the data of the multiple principal component belong to data under normal environment Confidence level.

In one example, the distribution of the data based at least one principal component and default confidence level determine exception ring Under border the data of at least one principal component may include: characteristic value based at least one principal component, it is described at least The freedom degree of the data of one principal component and the default confidence level, determine the confidence interval under the default confidence level；It is based on The data for falling at least one principal component outside the confidence interval determine at least one described principal component under abnormal environment Data.

On this basis, the data that default confidence level is not understood as multiple principal components can also be fallen in into above-mentioned confidence interval Confidence level.Confidence level is often indicated with percentage namely confidence level can be indicated with percentage.

In the first embodiment of above-mentioned example, it is assumed that at least one described principal component includes first principal component and Two principal components, the first principal component are the maximum principal components of characteristic value in the multiple principal component, and the Second principal component, is The principal component that the size of characteristic value is number two in the multiple principal component, the then feature based at least one principal component The freedom degree and the default confidence level of value, the data of at least one principal component, determine setting under the default confidence level Believe section, can specifically include following sub-step:

The two-dimentional scatterplot point of the data of sub-step 1062, the data for drawing the first principal component and the Second principal component, Butut, the two dimension scatter diagram is using the first principal component and Second principal component, as the reference axis of cartesian coordinate system.

As shown in Fig. 2, can be using first principal component as the abscissa of cartesian coordinate system, Second principal component, is Descartes's seat The ordinate for marking system, draws the two-dimentional scatter diagram of the data of first principal component and the data of Second principal component,.In Fig. 2, The coordinate value of one point can be indicated with (value of first principal component, the value of Second principal component).

Sub-step 1064, the characteristic value based on the first principal component, the freedom degree of the data of the first principal component and The default confidence level determines the long axis of fiducial confidence ellipse, and the characteristic value based on the Second principal component, the Second principal component, Data freedom degree and the default confidence level, determine the short axle of the fiducial confidence ellipse, wherein the fiducial confidence ellipse be used for table Levy the confidence interval.

For example, the long axis of fiducial confidence ellipse and short axle can be calculated respectively by following two formula:

Long axis=sqrt (s*A)

Short axle=sqrt (s*B)

Wherein, radical sign operation is opened in " sqrt " expression, and A is the characteristic value of first principal component, and B is the feature of Second principal component, Value；S=t1*t2*finv (p, t1, t3)/(t2), wherein " finv " indicates the inverse function of F distribution；P indicates default confidence level； T1 is the molecular freedom of finv, herein due to being to determine confidence interval for two principal components, t1=2；T2 is first The quantity or t2 of the data of principal component are the quantity of the data of Second principal component, and ginseng sees the above table 2 it is found that t2=n；T3 is finv Denominator freedom degree, t3=t2-1 in this example embodiment.

More specifically, it is assumed that n=2223, p=95%, t2=2223, then t3=2222, then accordingly:

S=2*2223*finv (0.95,2,2222)/(2222)

Sub-step 1066 is based on the long axis and the short axle, in the two-dimentional scatter diagram, centered on origin Draw the fiducial confidence ellipse.

Specifically as shown in Fig. 2, can be drawn different size of ellipse centered on origin under the default confidence levels of difference Circle.In Fig. 2, ellipse 10 to ellipse 50 is that default confidence level is equal to 95%, 99%, 99.9%, 99.99% and respectively Fiducial confidence ellipse when 99.999%.And in the example shown in Fig. 2, the long axis of fiducial confidence ellipse coordinate corresponding with first principal component Axis is parallel, and the short axle of fiducial confidence ellipse reference axis corresponding with Second principal component, is parallel.

On this basis, the data based at least one principal component described in falling in outside the confidence interval determine abnormal It may include: based on being located at institute in the two-dimentional scatter diagram under environment the step of the data of at least one principal component The point outside fiducial confidence ellipse is stated, determines the data of the first principal component under abnormal environment and the data of the Second principal component,.

In a kind of more specifically embodiment, institute can be located at from any quadrant of the two-dimentional scatter diagram It states and is selected in the point outside fiducial confidence ellipse a bit, as the data point under the abnormal environment；It is put based on the data in the flute Coordinate value in karr coordinate system determines the data of the first principal component and the Second principal component, under the abnormal environment Data.Certainly, in practical applications, can according to actual needs, selected according to the specific embodiment be more in it is different Data point under normal environment, so that it is determined that the data of the data of more first principal components and Second principal component, out.

In another more specifically embodiment, it can will be located in any quadrant of the two-dimentional scatter diagram The mass center of point outside the fiducial confidence ellipse, as the data point under the abnormal environment；It is put based on the data in the flute card Coordinate value in your coordinate system, determines the number of the data of the first principal component and the Second principal component, under the abnormal environment According to.In this way, the data point under can determining an abnormal environment respectively in four quadrants.

For example, in Fig. 2, when fiducial confidence ellipse is ellipse 10, can by the mass center 11 of the point in first quartile outside ellipse 10, It is oval in the mass center 13 and third quadrant of point in the mass center 12 of point, third quadrant in second quadrant outside ellipse 10 outside ellipse 10 One or more of mass center 14 of point outside 10, as the data point under abnormal environment.It, can when fiducial confidence ellipse is ellipse 20 It will be ellipse in the mass center 22 of the point in the mass center 21 of the point in first quartile outside ellipse 20, the second quadrant outside ellipse 20, third quadrant One or more of the mass center 24 of point in the mass center 23 and third quadrant of point outside circle 20 outside ellipse 20, as abnormal environment Under data point.And so on, it, can be by mass center 31, the mass center 32, mass center 33 outside ellipse 30 when fiducial confidence ellipse is ellipse 30 One or more of with mass center 34, as the data point under abnormal environment；It, can will be oval when fiducial confidence ellipse is ellipse 40 One or more of mass center 41, mass center 42, mass center 43 and mass center 44 outside 40, as the data point under abnormal environment；Alternatively, It, can be by one or more in mass center 51, mass center 52, mass center 53 and the mass center 54 outside ellipse 50 when fiducial confidence ellipse is ellipse 50 It is a, as the data point under abnormal environment.

It can be appreciated that since mass center is able to reflect the general trend and average level of multiple data, by any quadrant In be located at the mass center of point outside ellipse, first principal component under the abnormal environment acquired as the data point under abnormal environment and the The data of two principal components are more reliable, more reasonable.

Since in Fig. 2, the coordinate value of a point can be indicated with (value of first principal component, the value of Second principal component), It therefore, can be true by the abscissa value of data point in two kinds of specific embodiments of the data point under above-mentioned determining abnormal environment It is set to the data of first principal component under abnormal environment, the ordinate value of data point is determined as Second principal component, under abnormal environment Data.

Certainly, in addition to the scatter diagram using two principal components (first principal component and Second principal component), confidence is drawn Ellipse determines outside the data of the two principal components under abnormal environment, above-mentioned based at least in second of specific embodiment The distribution of the data of one principal component and default confidence level determine the data of at least one principal component under abnormal environment, May include: the distribution and default confidence level of the data based on first principal component, determine under abnormal environment described first it is main at The data divided；Wherein, the first principal component is the maximum principal component of characteristic value in the multiple principal component.Namely it is based on one Principal component determines confidence interval.

Specifically, can characteristic value, the default confidence level, the data of the first principal component based on first principal component Freedom degree, determine the confidence interval under the default confidence level；Data and the confidence area based on the first principal component Between, determine the data of the first principal component under abnormal environment.

At this point, determine confidence interval due to only having chosen first principal component, thus determine that confidence interval can be one One section of line segment on axis is tieed up, and this line segment is centered on the origin of the one-dimensional axis.It can be based on falling in above-mentioned line in this way Section exterior point determines the data of the first principal component under abnormal environment.

Alternatively, in the third specific embodiment, the distribution of the above-mentioned data based at least one principal component and pre- Reliability is set, determines the data of at least one principal component under abnormal environment, may include: the number based on first principal component According to, the distribution and default confidence level of the data of Second principal component, and third principal component, determine under abnormal environment that described first is main The data of the data of ingredient, the data of Second principal component, and third principal component；Wherein, the first principal component is the multiple master The maximum principal component of characteristic value in ingredient, the Second principal component, are that the size of characteristic value in the multiple principal component is number two Principal component, the Second principal component, is the principal component that the size of characteristic value in the multiple principal component is number three.Namely Confidence interval is determined based on three principal components.

Specifically, can be the coordinate for cartesian coordinate system with first principal component, Second principal component, and third principal component Axis draws the three-dimensional scatter diagram of the data of first principal component, the data of Second principal component, and third principal component；Described three It ties up in scatter diagram, the confidence ellipsoid under the default confidence level is drawn centered on origin, wherein the fiducial confidence ellipse The determination method of three half shaft lengths is similar with the method for the long axis of fiducial confidence ellipse identified above and short axle, does not do repetition herein and retouches It states；It is then based on the point being located at outside the confidence ellipsoid in the three-dimensional scatter diagram, determines that first under abnormal environment is main The data of the data of ingredient, the data of Second principal component, and third principal component.

It is conceivable that in step 106, the quantity for the principal component for including in " at least one principal component " is more, step The confidence interval determined in rapid 108 is reasonable, so that the achievement data under the abnormal environment predicted is more reliable.Certainly, described The quantity of " at least one principal component " is more, determines that the process of confidence interval is also more complicated.It in practical applications can be to this Two aspects make balance, determine the quantity for the principal component for including in suitable " at least one principal component ".

Step 108, to the data of at least one principal component described under the abnormal environment, successively carry out the principal component The inverse operation of analysis and the pretreated inverse operation, generate prediction data of the multiple index under the abnormal environment.

It is appreciated that the data of principal component are not necessarily referring to target data, need to obtain number of principal components according to reversed reduction is carried out The data of index.For example, the inverse transformation of the matrixing used when first with principal component analysis obtains the target data of index, so The target data of index is denormalized afterwards and changing value restores, obtains the prediction data of index.

Achievement data generation method under the abnormal environment that this specification embodiment provides, due to the target to multiple indexs Data carry out the data for multiple principal components that principal component analysis obtains, the connection being able to reflect between the historical data of multiple indexs Therefore system when using achievement data under principal component analysis predicted anomaly environment, can take into account the connection between different indexs System, improves the reliability and reasonability of the multiple achievement datas predicted, when carrying out pressure test using the achievement data, obtains Pressure testing results it is also more accurate.

In addition, compared to conventional method, achievement data generation method under the abnormal environment that this specification embodiment provides, With theoretical foundation, interpretation is strong, therefore the data predicted are more reliable, more reasonable.

Studies have shown that in the above-described embodiments, if different maturity periods multiple indexs are 8 kinds structure Shibor interest rate, on The maximum first principal component of characteristic value is able to reflect term structure level in text, and second largest Second principal component, of characteristic value can be anti- Term structure slope variation is reflected, it therefore, in one example, can be by step 106, to principal components multiple under abnormal environment The simulation and forecast of data is simplified to the simulation and forecast of the data to first principal component and Second principal component, under abnormal environment.

More specifically, wherein first principal component can reflect the floating and downlink of the Shibor interest rate of 8 kinds of term structures Situation implies that the Shibor interest rate of 8 kinds of term structures floats, otherwise implies 8 kinds of phases if first principal component is positive value The Shibor interest rate downlink of limit structure；Different maturity periods that Second principal component, can reflect structure Shibor interest rate between slope, If Second principal component, is positive value, gap different maturity periods showing between the Shibor interest rate of structure increases, otherwise shows difference Gap between the Shibor interest rate of term structure reduces.After the influence of two kinds of principal components is superimposed, just having can The abnormal conditions that such as interest rate is projecting can be predicted, so-called interest rate is projecting to refer to that the Shibor interest rate limited in short term is greater than long expiration The phenomenon that Shibor interest rate.

Fig. 3 to Fig. 6 shows the change of the Shibor interest rate of 8 kinds of term structures and the value of first principal component and Second principal component, Change relationship, in Fig. 3 into Fig. 6, what appended drawing reference 31 to 38 respectively corresponded expression be it is overnight, one week, two weeks, 1 month, 3 months, The Shibor interest rate of 6 months, 9 months and 1 year 8 kinds of term structure, and what Fig. 3 to Fig. 6 respectively indicated is that first quartile is extremely in Fig. 2 The corresponding Shibor change of interest rate situation of fourth quadrant.

Specifically, as shown in Figure 3 and Figure 6, when the value of first principal component is greater than 0, the Shibor interest rate of 8 kinds of term structures It is in rising trend；As shown in Figure 4 and Figure 5, when the value of first principal component is less than 0, the Shibor interest rate of 8 kinds of term structures is equal It is on a declining curve；Shown in Fig. 3 and Fig. 4, when the value of Second principal component, is greater than 0, different maturity periods structure Shibor interest rate between Gap increases；Shown in Fig. 5 and Fig. 6, when the value of Second principal component, is less than 0, different maturity periods structure Shibor interest rate between difference Away from reduction.And from fig. 6 it can be seen that the Shibor interest rate of 8 kinds of term structures is whole when the value of first principal component is greater than 0 Rise, when the value of Second principal component, is less than 0, different maturity periods structure Shibor interest rate between gap reduce, and occur The projecting extreme case of interest rate.From figure 3, it can be seen that when the value of first principal component is less than 0,8 kinds of term structures Shibor interest rate entire lowering, when the value of Second principal component, be greater than 0 when, different maturity periods structure Shibor interest rate between gap Increase, the projecting extreme case of interest rate also occurs.This meets the extreme feelings of the money market mobility mutation occurred in history Condition, therefore, the achievement data that the method that this specification provides predicts are more reasonable, more reliably.

Optionally, in another embodiment, on the basis of example shown in Fig. 2, as shown in fig. 7, in above-mentioned steps 108 Before, this specification embodiment provide abnormal environment under achievement data generation method, can also include:

Step 110, the data for determining remaining corresponding principal component of point in quadrant where the data point, it is described remaining Principal component is the principal component in the multiple principal component in addition to the first principal component and the Second principal component,.

For example, it is assumed that the multiple principal components determined in step 104 are gone back in addition to including first principal component and Second principal component, Including third principal component and the 4th principal component, then in step 110 it needs to be determined that the point in four quadrants shown in Fig. 2 corresponds to out Third principal component and the 4th principal component data.

Step 112, the standard deviation of data based on remaining principal component are determining the data of remaining principal component just State distribution map.

In this step, the corresponding third principal component of point and the 4th in determine in step 110 four quadrants is led The data of ingredient press the difference of corresponding quadrant respectively: first calculating the standard deviation of the data of third principal component, and calculate the 4th The standard deviation of the data of principal component；Then, the standard deviation of the data based on third principal component draws the data of third principal component Normal distribution, the standard deviation of the data based on the 4th principal component draw the normal distribution of the data of the 4th principal component.Finally It obtains in different quadrants, the normal distribution of the data of the normal distribution of the data of third principal component and the 4th principal component.

Step 114 is based on the normal distribution, determines the data of remaining principal component under abnormal environment.

Normal distribution can reflect the regularity of distribution of random sample, as shown in figure 8, normal distribution curve is a centre Height, both ends are gradually reduced and full symmetric bell curve, it is however generally that, the sample at the center far from bell curve occurs Probability it is smaller, and the sample under usually some abnormal environments.It therefore, can will be remote in the normal distribution of remaining principal component Data from bell curve center are determined as the data under abnormal environment, for example, by appended drawing reference 81 in Fig. 8 or 82 signified positions The data set, the data being determined as under abnormal environment.

As shown in fig. 7, at this time step 108 may include: under the abnormal environment, the data of the first principal component, The data of the data of the Second principal component, and remaining principal component, successively carry out the principal component analysis inverse operation and The pretreated inverse operation generates prediction data of the multiple index under the abnormal environment.

Specifically can be under to the abnormal environment, the data of the data of the first principal component, the Second principal component, And the data of remaining principal component merge and then successively carry out the inverse operation of the principal component analysis and described pre- The inverse operation of processing generates prediction data of the multiple index under the abnormal environment.

For example, it is assumed that the data point under the abnormal environment determined in step 106 is, two dimension scatter plot shown in Fig. 2 In first quartile in mass center 11, and assume mass center 11 coordinate be (1,2)；It determines under abnormal environment in step 114 Remaining principal component value be (0.1,0.2,0.3,0.4,0.5,0.6)；Can merge to obtain such one group (1,2,0.1,0.2, 0.3,0.4,0.5,0.6) then data execute the inverse behaviour of principal component analysis to (1,2,0.1,0.2,0.3,0.4,0.5,0.6) Work and pretreated inverse operation, generate prediction data of multiple indexs under abnormal environment.For other quadrants shown in Fig. 2 In mass center, can be handled using similar method, be not repeated to describe herein.

It can be appreciated that the achievement data under the abnormal environment that embodiment shown in Fig. 7 provides determines method, in addition to abnormal ring Outside the data of first principal component under border and the data of Second principal component, also the data of remaining principal component are merged, it is inverse To the data for predicting multiple indexs, therefore, the reliability and reasonability of the multiple achievement datas predicted can be further proposed.

It is that a kind of explanation of the achievement data generation method under abnormal environment is provided this specification above, below to this theory The electronic equipment that bright book provides is introduced.

Fig. 9 is the structural schematic diagram for the electronic equipment that one embodiment of this specification provides.Referring to FIG. 9, in hardware Level, the electronic equipment include processor, optionally further comprising internal bus, network interface, memory.Wherein, memory can It can include memory, such as high-speed random access memory (Random-Access Memory, RAM), it is also possible to further include non-easy The property lost memory (non-volatile memory), for example, at least 1 magnetic disk storage etc..Certainly, which is also possible to Including hardware required for other business.

Processor, network interface and memory can be connected with each other by internal bus, which can be ISA (Industry Standard Architecture, industry standard architecture) bus, PCI (Peripheral Component Interconnect, Peripheral Component Interconnect standard) bus or EISA (Extended Industry Standard Architecture, expanding the industrial standard structure) bus etc..The bus can be divided into address bus, data/address bus, control always Line etc..Only to be indicated with a four-headed arrow in Fig. 9, it is not intended that an only bus or a type of convenient for indicating Bus.

Memory, for storing program.Specifically, program may include program code, and said program code includes calculating Machine operational order.Memory may include memory and nonvolatile memory, and provide instruction and data to processor.

Processor is from the then operation into memory of corresponding computer program is read in nonvolatile memory, in logical layer The achievement data generating means under abnormal environment are formed on face.Processor executes the program that memory is stored, and is specifically used for Execute following operation:

The achievement data generation method under abnormal environment disclosed in the above-mentioned embodiment illustrated in fig. 1 such as this specification can be applied It is realized in processor, or by processor.Processor may be a kind of IC chip, the processing capacity with signal. During realization, each step of the above method can pass through the integrated logic circuit or software form of the hardware in processor Instruction complete.Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit, CPU), network processing unit (Network Processor, NP) etc.；It can also be digital signal processor (Digital Signal Processor, DSP), specific integrated circuit (Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic Device, discrete gate or transistor logic, discrete hardware components.It may be implemented or execute this specification one or more Disclosed each method, step and logic diagram in embodiment.General processor can be microprocessor or the processor It can be any conventional processor etc..The step of method in conjunction with disclosed in this specification one or more embodiment, can be straight Connect and be presented as that hardware decoding processor executes completion, or in decoding processor hardware and software module combination executed At.Software module can be located at random access memory, and flash memory, read-only memory, programmable read only memory or electrically-erasable can In the storage medium of this fields such as programmable memory, register maturation.The storage medium is located at memory, and processor reads storage Information in device, in conjunction with the step of its hardware completion above method.

The electronic equipment can also carry out the achievement data generation method under the abnormal environment of Fig. 1, and this specification is herein no longer It repeats.

Certainly, other than software realization mode, other implementations are not precluded in the electronic equipment of this specification, such as Logical device or the mode of software and hardware combining etc., that is to say, that the executing subject of following process flow is not limited to each Logic unit is also possible to hardware or logical device.

This specification embodiment also proposed a kind of computer readable storage medium, the computer-readable recording medium storage One or more programs, the one or more program include instruction, and the instruction is when by the portable electric including multiple application programs When sub- equipment executes, the method that the portable electronic device can be made to execute embodiment illustrated in fig. 1, and be specifically used for executing following Operation:

Achievement data generating means under a kind of abnormal environment provided below this specification are illustrated.

Figure 10 is the structural schematic diagram of the achievement data generating means 1000 under the abnormal environment that this specification provides.It please join Figure 10 is examined, a kind of achievement data generating means 1000 in Software Implementation, under abnormal environment can include: preprocessing module 1001, principal component analysis module 1002, the first determining module 1003 and generation module 1004.

Preprocessing module 1001 pre-processes respectively for the historical data to multiple indexs, obtains the multiple finger Target target data, the target data are the standardized data of the changing value between the historical data of different moments.

Optionally, preprocessing module 1001 specifically can be used for the historical data to the multiple index, respectively with default Time interval is changing value of the posterior data of step size computation generation time relative to the preceding data of generation time, is obtained described The variation Value Data of multiple indexs.

The variation Value Data of the multiple index is standardized respectively, obtains the number of targets of the multiple index According to

Principal component analysis module 1002 carries out principal component analysis for the target data to the multiple index, obtains more The data of a principal component.

First determining module 1003, for the distribution and default confidence level of the data based at least one principal component, really Determine the data of at least one principal component under abnormal environment, the default confidence level is that the data of the multiple principal component belong to The confidence level of data under normal environment.

Optionally, the first determining module 1003 specifically can be used for characteristic value based at least one principal component, institute The freedom degree and the default confidence level for stating the data of at least one principal component, determine the confidence area under the default confidence level Between；Based on the data of at least one principal component described in falling in outside the confidence interval, determine described at least one under abnormal environment The data of a principal component.

Optionally, in a kind of way of example, at least one above-mentioned principal component include first principal component and second it is main at Point, the first principal component is the maximum principal component of characteristic value in the multiple principal component, and the Second principal component, is described more The principal component that the size of characteristic value is number two in a principal component.And first determining module 1003 can be used for: draw described the The two-dimentional scatter diagram of the data of one principal component and the data of the Second principal component, the two dimension scatter diagram is with described First principal component and Second principal component, are the reference axis of cartesian coordinate system；It is characteristic value based on the first principal component, described The freedom degree of the data of first principal component and the default confidence level determine the long axis of fiducial confidence ellipse, and main based on described second The characteristic value of ingredient, the freedom degree of the data of the Second principal component, and the default confidence level, determine the fiducial confidence ellipse Short axle, wherein the fiducial confidence ellipse is for characterizing the confidence interval；Based on the long axis and the short axle, in the two dimension In scatter diagram, the fiducial confidence ellipse is drawn centered on origin.

And first determining module 1003 can be used for: based on being located at the fiducial confidence ellipse in the two-dimentional scatter diagram Outer point determines the data of the first principal component under abnormal environment and the data of the Second principal component,.

More specifically, the first determining module 1003 can be used for: from any quadrant of the two-dimentional scatter diagram It is selected in the point outside the fiducial confidence ellipse a bit, as the data point under the abnormal environment；It puts based on the data Coordinate value in the cartesian coordinate system determines the data of the first principal component and second master under the abnormal environment The data of ingredient.

Alternatively, the first determining module 1003 can be used for: institute will be located in any quadrant of the two-dimentional scatter diagram The mass center for stating the point outside fiducial confidence ellipse, as the data point under the abnormal environment；

The coordinate value in the cartesian coordinate system is put based on the data, is determined described first under the abnormal environment The data of the data of principal component and the Second principal component,.

Generation module 1004 is successively carried out for the data at least one principal component described under the abnormal environment The inverse operation of the principal component analysis and the pretreated inverse operation, generate the multiple index under the abnormal environment Prediction data.

Achievement data generating means 1000 under a kind of abnormal environment provided in this embodiment, due to the mesh to multiple indexs Mark data carry out the data for multiple principal components that principal component analysis obtains, the connection being able to reflect between the historical data of multiple indexs Therefore system when using achievement data under principal component analysis predicted anomaly environment, can take into account the connection between different indexs System, improves the reliability and reasonability of the multiple achievement datas predicted, when carrying out pressure test using the achievement data, obtains Pressure testing results it is also more accurate.

Achievement data generating means 1000 under the abnormal environment that another embodiment that Figure 11 implements this specification provides Structural schematic diagram, as shown in figure 11, the achievement data generating means 1000 under abnormal environment are in addition to including: preprocessing module 1001, principal component analysis module 1002, the first determining module 1003 and generation module 1004 can also include: the second determining mould Block 1005, third determining module 1006 and the 4th determining module 1007.

Second determining module 1005, for determining the quadrant where the data point before triggering generation module 1004 In remaining corresponding principal component of point data, remaining described principal component is in the multiple principal component except the first principal component With the principal component outside the Second principal component,.

Third determining module 1006 determines remaining described master for the standard deviation of the data based on remaining principal component The normal distribution of the data of ingredient.

4th determining module 1007 determines remaining described principal component under abnormal environment for being based on the normal distribution Data, and trigger generation module 1004.

And generation module 1004 is specifically used at this time: under the abnormal environment, the data of the first principal component, institute The data of Second principal component, and the data of remaining principal component are stated, inverse operation and the institute of the principal component analysis are successively carried out Pretreated inverse operation is stated, prediction data of the multiple index under the abnormal environment is generated.

Achievement data determining device 1000 under the abnormal environment that embodiment shown in Figure 11 provides, in addition under abnormal environment The data of first principal component and the data of Second principal component, outside, also the data of remaining principal component are merged, it is reverse pre- The data of multiple indexs are measured, therefore, can further propose the reliability and reasonability of the multiple achievement datas predicted.

It should be noted that the achievement data generating means 1000 under abnormal environment can be realized the embodiment of the method for Fig. 1 Method, specifically refer to the achievement data generation method under the abnormal environment of embodiment illustrated in fig. 1, repeat no more.

In short, being not intended to limit the protection of this specification the foregoing is merely the preferred embodiment of this specification Range.With within principle, made any modification, changes equivalent replacement all spirit in this specification one or more embodiment Into etc., it should be included within the protection scope of this specification one or more embodiment.

System, device, module or the unit that above-described embodiment illustrates can specifically realize by computer chip or entity, Or it is realized by the product with certain function.It is a kind of typically to realize that equipment is computer.Specifically, computer for example may be used Think personal computer, laptop computer, cellular phone, camera phone, smart phone, personal digital assistant, media play It is any in device, navigation equipment, electronic mail equipment, game console, tablet computer, wearable device or these equipment The combination of equipment.

Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data. The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM), Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.

It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want Element.When not limiting more, the element that is limited by sentence "including a ...", it is not excluded that in the mistake including the element There is also other identical elements in journey, method, commodity or equipment.

All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for system reality For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method Part explanation.

Claims

1. the achievement data generation method under a kind of abnormal environment, which comprises

The historical data of multiple indexs is pre-processed respectively, obtains the target data of the multiple index, the number of targets According to the standardized data of the changing value between the historical data for different moments；

The distribution of data based at least one principal component and default confidence level determine at least one described master under abnormal environment The data of ingredient, the default confidence level are that the data of the multiple principal component belong to the confidence level of data under normal environment；

To the data of at least one principal component described under the abnormal environment, the inverse operation of the principal component analysis is successively carried out With the pretreated inverse operation, prediction data of the multiple index under the abnormal environment is generated.

2. being obtained according to the method described in claim 1, wherein, the historical data to multiple indexs pre-processes respectively Obtain the target data of the multiple index, comprising:

To the historical data of the multiple index, respectively using prefixed time interval as the posterior data phase of step size computation generation time The changing value of data preceding for generation time obtains the variation Value Data of the multiple index；

The variation Value Data of the multiple index is standardized respectively, obtains the target data of the multiple index.

3. according to the method described in claim 1, wherein, the distribution of the data based at least one principal component and default Confidence level determines the data of at least one principal component under abnormal environment, comprising:

The freedom degree of the data of characteristic value, at least one principal component based at least one principal component and described default Confidence level determines the confidence interval under the default confidence level；

Based on the data of at least one principal component described in falling in outside the confidence interval, determine described at least one under abnormal environment The data of a principal component.

4. according to the method described in claim 3, at least one described principal component includes first principal component and Second principal component, institute Stating first principal component is the maximum principal component of characteristic value in the multiple principal component, the Second principal component, be it is the multiple it is main at The principal component that the size of characteristic value is number two in point；

Wherein, the characteristic value based at least one principal component, the freedom degree of the data of at least one principal component With the default confidence level, the confidence interval under the default confidence level is determined, comprising:

The two-dimentional scatter diagram of the data of the first principal component and the data of the Second principal component, is drawn, the two dimension dissipates Point distribution map is using the first principal component and Second principal component, as the reference axis of cartesian coordinate system；

The freedom degree and the default confidence level of the data of characteristic value, the first principal component based on the first principal component, Determine the long axis of fiducial confidence ellipse, and the freedom degree of the characteristic value based on the Second principal component, the data of the Second principal component, With the default confidence level, the short axle of the fiducial confidence ellipse is determined, wherein the fiducial confidence ellipse is for characterizing the confidence area Between；

It is ellipse that the confidence is drawn centered on origin in the two-dimentional scatter diagram based on the long axis and the short axle Circle.

It is described based at least one master described in falling in outside the confidence interval 5. according to the method described in claim 4, wherein The data of ingredient determine the data of at least one principal component under abnormal environment, comprising:

Based on the point being located at outside the fiducial confidence ellipse in the two-dimentional scatter diagram, determine that described first under abnormal environment is main The data of the data of ingredient and the Second principal component,.

6. described ellipse based on the confidence is located in the two-dimentional scatter diagram according to the method described in claim 5, wherein Point outside circle, determines the data of the data of the first principal component and the Second principal component, under abnormal environment, comprising:

It is selected a bit from the point being located at outside the fiducial confidence ellipse in any quadrant of the two-dimentional scatter diagram, as described Data point under abnormal environment；

Put the coordinate value in the cartesian coordinate system based on the data, determine under the abnormal environment described first it is main at The data of the data and the Second principal component, divided.

7. described ellipse based on the confidence is located in the two-dimentional scatter diagram according to the method described in claim 5, wherein Point outside circle, determines the data of the data of the first principal component and the Second principal component, under abnormal environment, comprising:

By the mass center for the point being located at outside the fiducial confidence ellipse in any quadrant of the two-dimentional scatter diagram, as the exception Data point under environment；

8. method according to claim 6 or 7, wherein it is described to the abnormal environment under described at least one master The data of ingredient successively carry out the inverse operation and the pretreated inverse operation of the principal component analysis, generate the multiple finger It is marked on before the prediction data under the abnormal environment, the method also includes:

Determine the data of remaining corresponding principal component of point in the quadrant where the data point, remaining described principal component is described Principal component in multiple principal components in addition to the first principal component and the Second principal component,；

The standard deviation of data based on remaining principal component determines the normal distribution of the data of remaining principal component；

Based on the normal distribution, the data of remaining principal component under abnormal environment are determined；

Wherein, the data at least one principal component described under the abnormal environment, successively carry out the principal component point The inverse operation of analysis and the pretreated inverse operation generate prediction data of the multiple index under the abnormal environment, packet It includes:

To under the abnormal environment, the data of the first principal component, the data of the Second principal component, and remaining described master The data of ingredient successively carry out the inverse operation and the pretreated inverse operation of the principal component analysis, generate the multiple finger The prediction data being marked under the abnormal environment.

9. the achievement data generating means under a kind of abnormal environment, described device include:

Preprocessing module pre-processes respectively for the historical data to multiple indexs, obtains the target of the multiple index Data, the target data are the standardized data of the changing value between the historical data of different moments；

Principal component analysis module carries out principal component analysis for the target data to the multiple index, obtains multiple principal components Data；

First determining module determines exception ring for the distribution and default confidence level of the data based at least one principal component The data of at least one principal component under border, the default confidence level is that the data of the multiple principal component belong to normal environment Under data confidence level；

Generation module, for the data at least one principal component described under the abnormal environment, successively carry out it is described it is main at The inverse operation and the pretreated inverse operation of analysis, generate prediction number of the multiple index under the abnormal environment According to.

10. a kind of electronic equipment, comprising:

Processor；And

It is arranged to the memory of storage computer executable instructions, the executable instruction makes the processor when executed Execute following operation:

The distribution of data based at least one principal component and default confidence level determine at least one described master under abnormal environment The data of ingredient, the default confidence level are the credible water that the data of the multiple principal component belong to the data under normal environment It is flat；

11. a kind of computer readable storage medium, the computer-readable recording medium storage one or more program, described one A or multiple programs are when the electronic equipment for being included multiple application programs executes, so that the electronic equipment executes following behaviour Make: