CN110134797A - A method of change of financial market is studied and judged based on reason map and multi-sector model - Google Patents
A method of change of financial market is studied and judged based on reason map and multi-sector model Download PDFInfo
- Publication number
- CN110134797A CN110134797A CN201910353532.6A CN201910353532A CN110134797A CN 110134797 A CN110134797 A CN 110134797A CN 201910353532 A CN201910353532 A CN 201910353532A CN 110134797 A CN110134797 A CN 110134797A
- Authority
- CN
- China
- Prior art keywords
- factor
- reason
- event
- current
- outlier
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/06—Asset management; Financial planning or analysis
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Computational Linguistics (AREA)
- Animal Behavior & Ethology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Technology Law (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of methods for studying and judging change of financial market based on reason map and multi-sector model, comprising: S1., which is obtained in historical events library, to be had causal various historical events and carry out polymerization evolution building reason map according to the influence relationship between each historical events；S2. match by result event and the historical events in reason map of the current event of acquisition, and extract and event evolution chain is generated to the various historical events in upstream that current event tool has an impact；S3. based on because word bank extract event evolution chain on directly or indirectly influence each historical events of current event the reason of the factor；S4. the reason of regular factor building in the reason factor influences current event variation factor set is chosen, calculates reason factor set to the explanation degree of the current results factor；If explanation degree is lower than explanation degree threshold value, the Outlier factor being introduced into the reason factor in reason factor set is until meet explanation degree threshold value.Of the invention studying and judging is high-efficient, and accuracy is high.
Description
Technical field
The present invention relates to information service field more particularly to a kind of methods for studying and judging change of financial market.
Background technique
Financial system is one of modern Economy Development mainstay, with networking, information-based fast development, in society
The event of generation, the propagation of the information in each face of national policy and each side are quick, and often these events can be to the change in financial market
Change has particularly important influence.For example, the fluctuation of staple commodities (futures or stock) price is highly susceptible to various events
Influence.These events include the publication of macro policy, supply variation, changes in demand, inventory change etc., but when different
Phase, it is different that the variation of each reason event, which brings the variation tendency of result event (such as forward price or spot price),
's.It is thorough that this Long-term change trend even investment research person also not necessarily considers.In the prior art, often through artificial right
The event of generation comprehensively consider the Long-term change trend in post analysis market, even if being considered various events, it is also difficult to fixed
A certain period reason event is depicted to the maximum factor of disturbance degree of result event (such as price) in amount, and especially a certain period goes out
Influence of the existing abnormal factors to turn of the market is difficult to by quantitative description.
Summary of the invention
The purpose of the present invention is to provide a kind of methods for studying and judging change of financial market, can intelligently become to financial market
Change carries out studying and judging early warning.
For achieving the above object, the present invention provides one kind and studies and judges financial market based on reason map and multi-sector model
The method of variation, comprising:
S1. obtaining in historical events library has causal various historical events and according between each historical events
Influence relationship carry out polymerization develop building reason map；
S2. match by result event and the historical events in the reason map of the current event of acquisition, and
It extracts and event evolution chain is generated to the various historical events in upstream that current event tool has an impact；
S3. based on because word bank extract directly or indirectly influenced on the event evolution chain current event it is each described in go through
The reason of historical event part the factor, wherein the reason factor be the historical events quantizating index；
S4. the reason of regular factor building in the reason factor influences current event variation factor set is chosen
It closes, and calculates the reason factor set to the explanation degree of the current results factor, wherein the current results factor is described works as
The quantizating index of preceding event；
If the reason factor set is lower than explanations degree threshold value to the explanation degree of the current results factor, described in calculating
Reason factor set is to the Outlier factor being introduced into the reason factor during the explanation degree of the current results factor until meeting
The explanation degree threshold value.
According to an aspect of the present invention, in step S3, will historical events corresponding with the reason factor in history
The number occurred in event base count and ranking, if the number that the historical events occurs meets preset regular event threshold
Being worth range, then the corresponding reason factor marker is regular factor, otherwise by the reason factor marker be it is abnormal because
Son.
According to an aspect of the present invention, in step S4, the regular factor building chosen in the reason factor influences institute
In the step of the reason of stating current event variation factor set, comprising:
S41. the historical results factor corresponding with the current event in history is obtained；
S42. the regular factor building reason factor set in the reason factor is chosen, and calculates the reason factor set
Close the explanation degree to the historical results factor；
S43. the reason factor set that explanation degree is higher than the explanation degree threshold value is obtained, and compares the original of acquisition
Because of the number of the reason factor in factor set, the least reason factor set of reason factor number and described current is chosen
As a result factor degree of explaining calculates.
According to an aspect of the present invention, in step S4, the reason factor set is calculated to the solution of the current results factor
In the step of degree of releasing, comprising:
S44. according to the explanation degree parameter of the current event definitions multivariate model；
S45. the reason factor set comprising regular factor is calculated to the current knot based on the multivariate model
The explanation degree of the fruit factor, and the corresponding reason factor set is calculated to institute based on the ranking of the historical events
State the history quantile of the explanation degree of the result factor；
S46. calculated explanation degree is compared with the explanation degree threshold value, if the explanation degree is lower than the explanation
Threshold value is spent, then introduces the Outlier factor into the reason factor set.
According to an aspect of the present invention, during calculating the reason factor set to the explanation degree of the current results factor
The Outlier factor being introduced into the reason factor is until in the step of meeting the explanation degree threshold value, further includes:
S47. using meet it is described explain degree threshold value as target, choose the Outlier factor and be added in the reason factor set
Construct new reason factor set；
S48. the new reason factor set is calculated to the explanation degree of the current results factor；
S49. step S47-S48 is repeated, the reason of explanation degree is greater than or equal to explanation degree threshold value factor set is obtained.
According to an aspect of the present invention, further includes:
S5. Outlier factor and the current results met in the reason factor set of the explanation degree threshold value are calculated
The correlation of the factor starts the monitoring Outlier factor if the correlation results obtained are more than preset threshold.
According to an aspect of the present invention, include: in step S5
S51. the Outlier factor is obtained in history between the historical results factor corresponding with the current event
Correlation, and obtain its history quantile；
S52. calculate the correlation of the Outlier factor with the current results factor one by one, and with the history quantile
It compares, obtains the Outlier factor more than 90% history quantile；
S53. the coefficient of the multivariate model is fitted using least square method, and based on the multivariable mould after fitting
Type is monitored the Outlier factor, wherein the multivariate model includes the routine in the new reason factor set
The factor, the Outlier factor of acquisition and the current results factor.
According to an aspect of the present invention, if the Outlier factor in the reason factor set is to the current results
The disturbance degree of factor variations becomes larger, then issues early warning；
The disturbance degree is explanation degree or correlation of the Outlier factor to the current results factor.
According to an aspect of the present invention, if occur in the reason factor set Outlier factor set to current results because
The case where explanation degree or correlation of son become larger, then retrieve historical factors data, if it exists the change of the Outlier factor in history
The case where changing the variation of the leading regular factor, then issue early warning.
According to an aspect of the present invention, when issuing early warning, by pre-warning signal quantization output into the reason map
The reason map is updated and is shown.
According to an aspect of the present invention, pre-warning signal quantization is exported into the reason map to the reason map
In the step of being updated and showing, modified in the reason map according to the numerical value of the regular factor or the Outlier factor
The size of node or color of expression event, and according to the Outlier factor and the correlation of the current results factor and its
Relationship relative to the 90% of the Outlier factor or 10% history quantile, which changes, indicates result thing in the reason map
The thickness or color of the directed line segment of correlation or other attributes between part and anomalous event.
According to an aspect of the present invention, in step S44, comprising:
S441. Estimation Optimization is carried out to the parameter of the multivariate model；
S442. the degree of fitting of the multivariate model is examined.
According to an aspect of the present invention, in step S46, the ranking based on the historical events calculates corresponding
In the step of reason factor set is to the history quantile of the explanation degree of the result factor, pass through one week time of setting
Phase, to the total history quantile calculated in the time cycle in history of the reason factor set.
According to an aspect of the present invention, further includes: when issuing early warning, then it is described because in word bank will with it is described current
The relevant Outlier factor of event is updated and saves.
According to an aspect of the present invention, it is based on the reason map construction search engine, by inputting key search
It the event evolution chain of dependent event and shows in history.
According to an aspect of the present invention, according to search result, the event evolution chain comprising selection is automatically generated
Study and judge report.
A kind of scheme according to the present invention had both featured each comprehensively by using the construction of event layers and quantization layer one
Kind upstream events avoid in by various news, report limited point to the variation relation (i.e. reason map) of downstream events
Analysis, and the change of correlation and its particular number between various factors can be quantitatively got by the calculating by quantization layer
Change.Work as in addition, can be treated according to the present invention by the statistical distribution of the historical factor for respectively causing upstream and downstream event change
Whether the numerical value of prefactor is in normal interval or abnormal section (such as higher or relatively low), and then can be accurately judged to
The validity of current factor pair current event variation can be especially accurately judged to the rare factor (i.e. Outlier factor) and current
The validity of event change.
Detailed description of the invention
Fig. 1 schematically shows a kind of step frame of the method for studying and judging change of financial market of embodiment according to the present invention
Figure；
Fig. 2 schematically shows a kind of operation knot of the method for studying and judging change of financial market of embodiment according to the present invention
Composition；
Fig. 3 schematically shows a kind of process of the method for studying and judging change of financial market of embodiment according to the present invention
Figure；
Fig. 4 schematically shows a kind of reason figure of the method for studying and judging change of financial market of embodiment according to the present invention
Compose structure figures；
Fig. 5 schematically shows a kind of event base of the method for studying and judging change of financial market of embodiment according to the present invention
Template；
Fig. 6 schematically shows that a kind of extraction of the method for studying and judging change of financial market of embodiment according to the present invention is worked as
The flow chart of the current event of preceding generation；
Fig. 7 schematically shows a kind of history thing of the method for studying and judging change of financial market of embodiment according to the present invention
The part factor and historical events factor fitting result figure.
Specific embodiment
It, below will be to embodiment in order to illustrate more clearly of embodiment of the present invention or technical solution in the prior art
Needed in attached drawing be briefly described.It should be evident that the accompanying drawings in the following description is only of the invention some
Embodiment for those of ordinary skills without creative efforts, can also be according to these
Attached drawing obtains other attached drawings.
When being described for embodiments of the present invention, term " longitudinal direction ", " transverse direction ", "upper", "lower", " preceding ",
" rear ", "left", "right", "vertical", "horizontal", "top", "bottom" "inner", orientation or positional relationship expressed by "outside" are based on phase
Orientation or positional relationship shown in the drawings is closed, is merely for convenience of description of the present invention and simplification of the description, rather than instruction or dark
Show that signified device or element must have a particular orientation, be constructed and operated in a specific orientation, therefore above-mentioned term cannot
It is interpreted as limitation of the present invention.
The present invention is described in detail with reference to the accompanying drawings and detailed description, embodiment cannot go to live in the household of one's in-laws on getting married one by one herein
It states, but therefore embodiments of the present invention are not defined in following implementation.
As shown in Figure 1, a kind of embodiment according to the present invention, one kind of the invention is based on reason map and multiple-factor mould
The method that type studies and judges change of financial market, comprising:
S1. obtaining in historical events library has causal various historical events and according to the shadow between each historical events
The relationship of sound carries out polymerization evolution building reason map；
S2. match by result event and the historical events in reason map of the current event of acquisition, and extract to working as
The various historical events in upstream that preceding event tool has an impact generate event evolution chain；
S3. based on because word bank extract event evolution chain on directly or indirectly influence each historical events of current event the reason of
The factor, wherein the reason factor is the quantizating index of historical events；
S4. the reason of regular factor building in the reason factor influences current event variation factor set is chosen, and is calculated
Explanation degree of the reason factor set to the current results factor, wherein the current results factor is the quantizating index of current event；
If reason factor set is lower than explanation degree threshold value to the explanation degree of the current results factor, reason factor set is calculated
To the Outlier factor being introduced into during the explanation degree of the current results factor in the reason factor until meeting explanation degree threshold value.
In conjunction with shown in Fig. 1, Fig. 2 and Fig. 3, a kind of embodiment according to the present invention in step S1, obtains historical events library
In have causal various historical events and according to the influence relationship between each historical events carry out polymerization evolution structure
Build reason map.In the present embodiment, firstly, grinding report to history according to default rule, news carries out parsing and obtains
The event (i.e. historical events) of generation；Then, polymerization evolution is carried out according to the correlation between historical events and is configured to large quotient
The reason map of product industrial chain core.It should be pointed out that mutual due to can also exist between different staple commodities
Influence relationship, and then include the Evolvement of a variety of staple commodities in the reason map of building.In the present embodiment, according to
The process with causal historical events building reason map of acquisition is also that definition node (indicating event) arrives definition side
The process (referring to fig. 2 and Fig. 4) of (indicating correlation between the factor of event).In the present embodiment, historical events can be
Verb-object word group (event of price change is such as represented, such as: " U.S. soybean yield increases by ten thousand tons of ten thousand tondal xx of xx ").In addition, going through
Historical event part may also include index group and time attribute (such as passing by, present, future, section).
In the present embodiment, the core index between historical events and event is subjected to the production of polymeric configuration staple commodities
The reason map of industry chain core, for example, entity relation between supply and demand: macroscopic view, supply, demand, inventory, processing profit, financial derivatives
Relation between supply and demand: the amount of holding position.In the present embodiment, in the step of grinding report parsing historical events from history, from the following aspects
It carries out, is respectively objective event (event such as the fact only statement has occurred), (such as: " SS is for the event with prediction property
TT event occurs through generation/expection generation/maximum probability generation/small probability " etc.), event trigger word (as " initiation ", " causing ", "
Bring " etc.) and one period statistics appear in the number ground in report.From the step of parsing historical events in news, from two
Aspect carries out, and is respectively critical event (usually subsequent report) and public sentiment (number that certain event is mentioned), passes through public sentiment
The event be can express in the temperature in this period.As shown in Fig. 2, event a, event b and event c are in the reason map of building
In the causal event that t moment occurs, wherein event a is represented: soybean price changing trend (rises or falls), and event b is represented:
Soybean yields changes (rising or falling), and event c is represented: soybean consumption figure changes (rising or falling), then by event to put
Form is indicated, and is indicated the causality between event with line segment and connected to constitute reason map.
A kind of embodiment according to the present invention, in step s 2, using the current event of acquisition as result event and reason
Historical events in map matches, and extracts and generate event to the various historical events in upstream that current event tool has an impact
Evolution chain.In the present embodiment, the historical events parsed in report and financial and economic news is ground from history in the form of event library template
Stored (shown in Figure 5).Shown in Figure 6, method of the invention is by NLP (natural language processing) engine according to instruction
Practice the model completed and grinds the index i.e. current event for parsing in report or financial and economic news and currently occurring to what is currently obtained.From event base
The keyword that middle acquisition matches with current event, so match historical events and current event (i.e. current event with match
Historical events be similar event), then according to reason map extract using current event as the event evolution chain of result event, should
Event evolution chain includes and the direct or indirect relevant upstream historical events of current event.It certainly, can also in the present processes
Current event is inputted in a manner of using and be manually entered.In the present embodiment, the event evolution chain of extraction is in reason map
Belong to the event evolution chain (i.e. same staple commodities) of same class event with current event.
It according to the present invention, can be fast and effective by using the method for natural language processing compared to the method being manually entered
Identify the event ground in the data such as report, news, guarantee that method of the invention can be in time intelligently to current financial city
Field variation carries out studying and judging early warning.
In conjunction with shown in Fig. 2 and Fig. 3, a kind of embodiment according to the present invention, in step S3, based on extracting event because of word bank
The reason of each historical events of current event is directly or indirectly influenced on evolution chain the factor, wherein the reason factor be history
Event quantizating index (for example, signified event is " U.S. soybean yield reaches ten thousand tons of xx within 2018 ", then quantizating index are as follows: " 2018
Year U.S. soybean yield ", the value of the index is " ten thousand tons of xx ").In the present embodiment, because word bank passes through the following aspects
It realizes, firstly, selection reasonable time scale (or time interval), wherein time scale can be kept away as unit of week, moon etc.
Exempt from daily fluctuation or noise, then, occurrence is gone out to historical events (reason) to historical events (result) according to grinding in report or news
Number collects statistics and ranking, extracts in contrast if the number that historical events occurs meets preset regular event threshold range
The quantizating index answered simultaneously is labeled as regular factor and initializes regular factor library；If the number that historical events occurs is unsatisfactory for presetting
Regular event threshold range then extract corresponding quantizating index and labeled as Outlier factor and structural anomaly is because of word bank,
Referring to fig. 2, in the present embodiment, because including regular factor library and Outlier factor library in word bank.Certainly, according to the present invention another
A kind of embodiment enumerates influence because word bank can also be identified and/or be ground report interpretive mode by expert during building
(its value is integer to the preceding n of the result factor (target elements) of historical events, such as 10) the big reason factor, and finds out and enumerate
The explanation degree of the reason factor pair result factor is more than that these factors of preset value are used for the initialization in regular factor library in total, remaining
Factor structure Outlier factor library.In the present embodiment, because being stored with a variety of causes of various time points in history in word bank
The factor (such as: the quantizating index fb of the event b as reason event, the quantizating index fc of the event c as reason event), knot
The fruit factor numerical value (such as: the quantizating index fa of event a), and with event a, event b and event c occur the moment multivariable
Model fitting parameter values (such as: β0、β1And β2)。
A kind of embodiment according to the present invention, in step S4, the regular factor building chosen in the reason factor influences to work as
In the step of the reason of preceding event change factor set, comprising:
S41. the historical results factor corresponding with current event in history is obtained.In the present embodiment, according to the factor
Library obtains the quantizating index (i.e. the historical results factor) to match in history with current event.
S42. the regular factor chosen in the reason factor constructs reason factor set, and calculates reason factor set to history
As a result the explanation degree of the factor.In the present embodiment, it during constructing reason factor set, is obtained not using Main cause analysis method
The explanation degree (or importance) of Main Factors set of the same period.For example, be once mentioned first, in accordance with historical reasons event
Number (from more to few) sorts, and the reason factor number in reason factor set can be according to from less (only comprising most important history
Event) it is chosen to the sequence of more (containing more but few some events being mentioned).Therefore, factor the reason of building
Collection is combined into multiple, calculates separately each reason factor set to the explanation degree of the historical results factor.
S43. in the reason of obtaining the reason of explanation degree is higher than explanation degree threshold value factor set, and comparing acquisition factor set
The reason of the factor number, choose the minimum reason factor set of reason factor number and close and current results factor degree of explaining meter
It calculates.
A kind of embodiment according to the present invention in step S4, calculates reason factor set to the solution of the current results factor
In the step of degree of releasing, comprising:
S44. according to the explanation degree parameter of current event definitions multivariate model.In the present embodiment, by building
Vertical multivariate model (multiple linear regression model can be used) is fitted the relationship between current event and the reason factor, and
Calculate its explanation degree.
In the present embodiment, comprising:
S441. multivariate model is established based on the current results factor (dependent variable) and reason factor set (independent variable).So
Afterwards, using the optimized parameter of least square method OLS estimation multivariate model.It is shown in Figure 2, it is drilled according to the event in reason figure
Change the Relation acquisition current results factor and reason factor set constructs multivariate model fa=β0+β1*fb+β2*fc+εa, wherein
Fa is the quantizating index (the current results factor) of event a, i.e. XX yuan/ton of t moment (price)；The quantizating index that fb is event b is (former
Because of the regular factor in factor set), i.e. XX tons of t moment (yield)；Fc is the quantizating index of event c (in reason factor set
Regular factor), i.e. XX tons of t moment (consumption figure), since the weight of the factor is time-varying, and factor beta0、β1And β2When different
Between fitted data be also different.εaWhat is represented is the residual error after fitting.
S442. the degree of fitting of multivariate model is examined.Common degree of fitting index be R squares of index (R-Square) and
More reasonably R squares of index (Adjusted R-Square) of correction.In the present embodiment, using the latter (referring to Fig. 7 institute
Show).It is defined as follows:
SST=SSreg+SSe
Wherein, SST indicates total sum of squares, SSregIndicate regression sum of square, SSeIndicate residual sum of squares (RSS).
By above formula Deformation partition, a module of models fitting degree be may be expressed as:
Adjusted R-Square
Wherein, n is sample size, and p is characterized quantity, i.e. sample is n, [xi1,xi2,xi3,…,xip,yi], wherein i
=1 ... n.Y in above-mentioned sampleiFor the embodiment of a historical result factor, xi1,xi2,xi3,…,xipIt is in history one
The set of the group reason factor.
S45. reason factor set is calculated to the explanation degree of the current results factor based on multivariate model, and is based on history
The ranking of event calculates the reason of corresponding factor set conjunction to the history quantile of the explanation degree of the result factor；
S46. calculated explanation degree is compared with explanation degree threshold value, if explanation degree is lower than explanation degree threshold value, to
Outlier factor is introduced in reason factor set.In the present embodiment, the ranking based on historical events calculates corresponding
In the step of reason factor set is to the history quantile of the explanation degree of the result factor, by setting a time cycle, to original
Because of the total history quantile calculated in the time cycle in history of factor set.
A kind of embodiment according to the present invention in step S4, calculates reason factor set to the solution of the current results factor
The Outlier factor in the reason factor is introduced into during degree of releasing until in the step of meeting explanation degree threshold value, further includes:
S47. degree threshold value is explained as target to meet, choose in Outlier factor addition reason factor set and construct new reason
Factor set.In the present embodiment, by Outlier factor is added to obtained in abovementioned steps S43 to the current results factor
New reason factor set is formed in the reason of degree of explaining calculates factor set.
S48. new reason factor set is calculated to the explanation degree of the current results factor.
S49. step S47-S48 is repeated, the reason of explanation degree is greater than or equal to explanation degree threshold value factor set is obtained.At this
In embodiment, the new reason factor set of acquisition is the set for including regular factor and Outlier factor.
A kind of embodiment according to the present invention, method of the invention further include: S5. calculates the original for meeting explanation degree threshold value
Because of the Outlier factor and the correlation of the current results factor in factor set, if the correlation results obtained are more than preset threshold,
Then start monitoring Outlier factor.In the present embodiment, comprising:
S51. correlation of the Outlier factor in history between the historical results factor corresponding with current event is obtained, and
Obtain its history quantile.
S52. the correlation of Outlier factor and the current results factor is calculated one by one, and is compared with history quantile, is obtained super
Cross the Outlier factor of 90% history quantile.
S53. using the coefficient of least square method fitting multivariate model, and based on the multivariate model after fitting to exception
The factor is monitored, wherein multivariate model include new reason factor set in regular factor, the Outlier factor of acquisition and
The current results factor.
A kind of embodiment according to the present invention, if the Outlier factor in reason factor set is to current results factor variations
Disturbance degree become larger, then issue early warning.In the present embodiment, disturbance degree is explanation degree of the Outlier factor to the current results factor
Or correlation.In the present embodiment, if occurring explanation of the Outlier factor set to the current results factor in reason factor set
The case where degree or correlation become larger, then retrieve historical factors data, if it exists the change of the leading regular factor of variation of Outlier factor
The case where change, then issues early warning.
A kind of embodiment according to the present invention, when issuing early warning, by pre-warning signal quantization output into reason map
Reason map is updated and is shown.In the present embodiment, pre-warning signal quantization is exported into reason map to reason
In the step of map is updated and shows, being modified in reason map according to the numerical value of regular factor or Outlier factor indicates event
Size of node or color.Meanwhile according to the correlation of Outlier factor and the current results factor corresponding thereto in the exception because
The relationship of 90% or 10% history quantile of son, which changes, indicates correlation between result event and anomalous event in reason map
Directed line segment thickness or color.In the present embodiment, principle is more to be upwardly deviated from 90% quantile, or be deflected downwardly
10% quantile then assigns bigger numerical value.
A kind of embodiment according to the present invention then will be relevant to current event in because of word bank when issuing early warning
Outlier factor is updated and saves.
As shown in figure 3, according to an embodiment of the present invention, being based on reason map construction search engine, closed by input
Key search words the event evolution chain of dependent event and is shown in history.In the present embodiment, can by oneself interested one
The appearance of similar mode and Evolvement on group dependent event search history.In the present embodiment, according to search result, certainly
Dynamic event evolution chain of the generation comprising selection studies and judges report.By above-mentioned setting, so that the present invention is implementing to study and judge current thing
It is convenient that the Evolvement of historical events is checked during part changes, it greatly improves to current and historic market
The comprehensive analysis efficiency of variation.
According to the method for the present invention, the number occurred in history according to historical events divides the corresponding reason factor
Class, and then during capable of being studied and judged to the event currently occurred, directly selection regular factor constructs reason factor set
It is studied and judged.If regular factor meets explanation degree threshold value to the explanation degree of current event in reason factor set, can judge
Current event is a kind of regular event out.If regular factor cannot expire the explanation degree of current event in conventional reason factor set
Foot explanation degree threshold value can then judge that current event is a kind of anomalous event.It by the above process can be accurate and quick
Screen out current event whether anomalous event, improve and of the invention study and judge accuracy and timeliness.
According to the method for the present invention, it is anomalous event when studying and judging out current event, then further passes through phase reason factor set
The method for introducing Outlier factor in conjunction quantitatively studies and judges influence of the Outlier factor to current event, and then can accurately obtain out and work as
The change direction of the maximally related Outlier factor of preceding event and current event.Therefore, method of the invention is realized to anomalous event
And the real-time monitoring and early warning of Outlier factor, it further improves to the comprehensive accurate of event monitoring.
Above content is only the example of concrete scheme of the invention, for the equipment and structure of wherein not detailed description, is answered
When being interpreted as that the existing common apparatus in this field and universal method is taken to be practiced.
The foregoing is merely a schemes of the invention, are not intended to restrict the invention, for the technology of this field
For personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (16)
1. a kind of method for studying and judging change of financial market based on reason map and multi-sector model, comprising:
S1. obtaining in historical events library has causal various historical events and according to the shadow between each historical events
The relationship of sound carries out polymerization evolution building reason map；
S2. match by result event and the historical events in the reason map of the current event of acquisition, and extract
Event evolution chain is generated to the various historical events in upstream that current event tool has an impact；
S3. based on directly or indirectly influencing each history thing of the current event on the event evolution chain because word bank extracts
The reason of part the factor, wherein the reason factor be the historical events quantizating index；
S4. the reason of regular factor building in the reason factor influences current event variation factor set is chosen, and
The reason factor set is calculated to the explanation degree of the current results factor, wherein the current results factor is the current thing
The quantizating index of part；
If the reason factor set is lower than explanation degree threshold value to the explanation degree of the current results factor, the reason is calculated
Factor set is to the Outlier factor being introduced into the reason factor during the explanation degree of the current results factor until described in satisfaction
Explanation degree threshold value.
2. the method according to claim 1, wherein corresponding with the reason factor will be gone through in step S3
The number that historical event part occurs in historical events library count and ranking, presets if the number that the historical events occurs meets
Regular event threshold range then the corresponding reason factor marker be regular factor, otherwise by the reason factor
Labeled as Outlier factor.
3. method according to claim 1 or 2, which is characterized in that in step S4, choose the routine in the reason factor
Factor building influenced in the step of the reason of current event variation factor set, comprising:
S41. the historical results factor corresponding with the current event in history is obtained；
S42. the regular factor building reason factor set in the reason factor is chosen, and calculates the reason factor set pair
The explanation degree of the historical results factor；
S43. obtain the reason factor set of the explanations degree higher than the explanation degree threshold value, and compare the reason of acquisition because
The number of the reason factor in subclass chooses the least reason factor set of reason factor number and the current results
Factor degree of explaining calculates.
4. according to the method described in claim 3, it is characterized in that, calculating the reason factor set to current in step S4
As a result in the step of explanation of the factor is spent, comprising:
S44. according to the explanation degree parameter of the current event definitions multivariate model；
S45. based on the multivariate model calculate comprising regular factor the reason factor set to the current results because
The explanation degree of son, and the corresponding reason factor set is calculated to the knot based on the ranking of the historical events
The history quantile of the explanation degree of the fruit factor；
S46. calculated explanation degree is compared with the explanation degree threshold value, if the explanation degree is lower than the explanation degree threshold
Value, then introduce the Outlier factor into the reason factor set.
5. according to the method described in claim 4, it is characterized in that, calculating the reason factor set to the current results factor
The Outlier factor being introduced into the reason factor during explanation degree also wraps until in the step of meeting the explanation degree threshold value
It includes:
S47. using meet it is described explain degree threshold value as target, choose the Outlier factor and be added in the reason factor set and construct
New reason factor set；
S48. the new reason factor set is calculated to the explanation degree of the current results factor；
S49. step S47-S48 is repeated, the reason of explanation degree is greater than or equal to explanation degree threshold value factor set is obtained.
6. according to the method described in claim 5, it is characterized by further comprising:
S5. Outlier factor and the current results factor met in the reason factor set of the explanation degree threshold value is calculated
Correlation start the monitoring Outlier factor if the correlation results obtained are more than preset threshold.
7. according to the method described in claim 6, it is characterized in that, including: in step S5
S51. it is related between the historical results factor corresponding to the current event in history to obtain the Outlier factor
Property, and obtain its history quantile；
S52. calculate the correlation of the Outlier factor with the current results factor one by one, and with the history quantile ratio
It is right, obtain the Outlier factor more than 90% history quantile；
S53. the coefficient of the multivariate model is fitted using least square method, and based on the multivariate model pair after fitting
The Outlier factor is monitored, wherein and the multivariate model includes the regular factor in the new reason factor set,
The Outlier factor obtained and the current results factor.
8. method according to claim 6 or 7, which is characterized in that if in the reason factor set it is described it is abnormal because
Son becomes larger to the disturbance degree of the current results factor variations, then issues early warning；
The disturbance degree is explanation degree or correlation of the Outlier factor to the current results factor.
9. according to the method described in claim 8, it is characterized in that, if there is Outlier factor set in the reason factor set
The case where becoming larger to the explanation degree or correlation of the current results factor, then retrieve historical factors data, described in history if it exists
The case where variation of the leading regular factor of variation of Outlier factor, then issue early warning.
10. according to the method described in claim 9, it is characterized in that, pre-warning signal quantization is exported to institute when issuing early warning
It states in reason map and the reason map is updated and is shown.
11. according to the method described in claim 10, it is characterized in that, by pre-warning signal quantization output into the reason map
In the step of reason map is updated and is shown, modified according to the numerical value of the regular factor or the Outlier factor
Indicate the size of node or color of event in the reason map, and according to the Outlier factor and the current results because
The correlation of son changes the reason figure in the relationship of the 90% of the Outlier factor or 10% history quantile corresponding thereto
The thickness or color of the directed line segment of correlation between result event and anomalous event are indicated in spectrum.
12. according to the method described in claim 4, it is characterized in that, in step S44, comprising:
S441. Estimation Optimization is carried out to the parameter of the multivariate model；
S442. the degree of fitting of the multivariate model is examined.
13. according to the method described in claim 4, it is characterized in that, in step S46, based on the ranking of the historical events
In the step of calculating history quantile of the corresponding reason factor set to the explanation degree of the result factor, pass through
A time cycle is set, to the total history quantile calculated in the time cycle in history of the reason factor set.
14. according to the method described in claim 8, it is characterized by further comprising: when issuing early warning, then described because of word bank
In the Outlier factor relevant to the current event is updated and is saved.
15. method described in 0 or 14 according to claim 1, which is characterized in that it is based on the reason map construction search engine,
It the event evolution chain of dependent event and is shown in history by input key search.
16. according to the method for claim 15, which is characterized in that according to search result, automatically generate the institute comprising choosing
That states event evolution chain studies and judges report.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910353532.6A CN110134797A (en) | 2019-04-29 | 2019-04-29 | A method of change of financial market is studied and judged based on reason map and multi-sector model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910353532.6A CN110134797A (en) | 2019-04-29 | 2019-04-29 | A method of change of financial market is studied and judged based on reason map and multi-sector model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110134797A true CN110134797A (en) | 2019-08-16 |
Family
ID=67575672
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910353532.6A Pending CN110134797A (en) | 2019-04-29 | 2019-04-29 | A method of change of financial market is studied and judged based on reason map and multi-sector model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110134797A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110807104A (en) * | 2019-11-08 | 2020-02-18 | 上海秒针网络科技有限公司 | Method and device for determining abnormal information, storage medium and electronic device |
| CN110968699A (en) * | 2019-11-01 | 2020-04-07 | 数地科技(北京)有限公司 | Logic map construction and early warning method and device based on event recommendation |
| CN111383102A (en) * | 2020-03-27 | 2020-07-07 | 北京明略软件系统有限公司 | Financial credit risk identification method, model construction method and device |
| CN111738532A (en) * | 2020-08-14 | 2020-10-02 | 支付宝(杭州)信息技术有限公司 | Method and system for acquiring influence degree of event on object |
| CN112948552A (en) * | 2021-02-26 | 2021-06-11 | 北京信息科技大学 | Method and device for online expansion of affair map |
| CN113449116A (en) * | 2021-06-22 | 2021-09-28 | 青岛海信网络科技股份有限公司 | Map construction and early warning method, device and medium |
| CN114846460A (en) * | 2019-12-23 | 2022-08-02 | 艾玛迪斯简易股份公司 | System and method for optimizing the transmission of requests for updated content from external data sources |
| US11789932B2 (en) | 2019-12-23 | 2023-10-17 | Amadeus S.A.S. | System and method for optimizing transmission of requests for updated content from external data sources |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107292097A (en) * | 2017-06-14 | 2017-10-24 | 华东理工大学 | The feature selection approach of feature based group and traditional Chinese medical science primary symptom system of selection |
| CN108052576A (en) * | 2017-12-08 | 2018-05-18 | 国家计算机网络与信息安全管理中心 | A kind of reason knowledge mapping construction method and system |
| CN108171602A (en) * | 2017-06-28 | 2018-06-15 | 永辉青禾商业保理(重庆)有限公司 | The method for building up and device of accounts receivable factoring business risk model |
| CN108876106A (en) * | 2018-05-14 | 2018-11-23 | 华南理工大学 | A kind of macro Micro dynamic mixing β estimation method merging macro economic policy factor |
| US20190012252A1 (en) * | 2013-08-01 | 2019-01-10 | Alpha Beta Analytics, LLC | Systems and methods for automated analysis, screening, and reporting of group performance |
| CN109242691A (en) * | 2018-08-08 | 2019-01-18 | 韩岩 | For the quantificational description method and device of the market data fluctuations of financial product |
-
2019
- 2019-04-29 CN CN201910353532.6A patent/CN110134797A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190012252A1 (en) * | 2013-08-01 | 2019-01-10 | Alpha Beta Analytics, LLC | Systems and methods for automated analysis, screening, and reporting of group performance |
| CN107292097A (en) * | 2017-06-14 | 2017-10-24 | 华东理工大学 | The feature selection approach of feature based group and traditional Chinese medical science primary symptom system of selection |
| CN108171602A (en) * | 2017-06-28 | 2018-06-15 | 永辉青禾商业保理(重庆)有限公司 | The method for building up and device of accounts receivable factoring business risk model |
| CN108052576A (en) * | 2017-12-08 | 2018-05-18 | 国家计算机网络与信息安全管理中心 | A kind of reason knowledge mapping construction method and system |
| CN108876106A (en) * | 2018-05-14 | 2018-11-23 | 华南理工大学 | A kind of macro Micro dynamic mixing β estimation method merging macro economic policy factor |
| CN109242691A (en) * | 2018-08-08 | 2019-01-18 | 韩岩 | For the quantificational description method and device of the market data fluctuations of financial product |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110968699A (en) * | 2019-11-01 | 2020-04-07 | 数地科技(北京)有限公司 | Logic map construction and early warning method and device based on event recommendation |
| CN110968699B (en) * | 2019-11-01 | 2023-07-18 | 数地工场(南京)科技有限公司 | Logic map construction and early warning method and device based on fact recommendation |
| CN110807104A (en) * | 2019-11-08 | 2020-02-18 | 上海秒针网络科技有限公司 | Method and device for determining abnormal information, storage medium and electronic device |
| CN114846460A (en) * | 2019-12-23 | 2022-08-02 | 艾玛迪斯简易股份公司 | System and method for optimizing the transmission of requests for updated content from external data sources |
| US11789932B2 (en) | 2019-12-23 | 2023-10-17 | Amadeus S.A.S. | System and method for optimizing transmission of requests for updated content from external data sources |
| CN111383102A (en) * | 2020-03-27 | 2020-07-07 | 北京明略软件系统有限公司 | Financial credit risk identification method, model construction method and device |
| CN111383102B (en) * | 2020-03-27 | 2023-10-24 | 北京明略软件系统有限公司 | Financial credit risk identification method, model construction method and device |
| CN111738532A (en) * | 2020-08-14 | 2020-10-02 | 支付宝(杭州)信息技术有限公司 | Method and system for acquiring influence degree of event on object |
| CN112948552A (en) * | 2021-02-26 | 2021-06-11 | 北京信息科技大学 | Method and device for online expansion of affair map |
| CN112948552B (en) * | 2021-02-26 | 2023-06-02 | 北京信息科技大学 | Online expansion method and device for a rational map |
| CN113449116A (en) * | 2021-06-22 | 2021-09-28 | 青岛海信网络科技股份有限公司 | Map construction and early warning method, device and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110134797A (en) | A method of change of financial market is studied and judged based on reason map and multi-sector model | |
| US20230213895A1 (en) | Method for Predicting Benchmark Value of Unit Equipment Based on XGBoost Algorithm and System thereof | |
| CN115796372B (en) | SCOR-based supply chain management optimization method and system | |
| US20120290497A1 (en) | Failure diagnosis system, failure diagnosis device and failure diagnosis program | |
| CN112446526B (en) | Production scheduling system and method | |
| CN112579640B (en) | Method and apparatus for production anomaly detection | |
| Ilangkumaran et al. | Application of hybrid VIKOR model in selection of maintenance strategy | |
| CN106067094A (en) | A kind of dynamic assessment method and system | |
| CN1494671A (en) | Data sharing in process plant | |
| CN103838202A (en) | Parameter control method and parameter control system | |
| CN108510209A (en) | A kind of process failure pattern-recognition and evaluation method based on fuzzy theory | |
| US20230152786A1 (en) | Industrial equipment operation, maintenance and optimization method and system based on complex network model | |
| CN106600310A (en) | Method and system for sales prediction based on network search index | |
| CN110826237B (en) | Wind power equipment reliability analysis method and device based on Bayesian belief network | |
| CN108491991A (en) | Constraints analysis system based on the industrial big data product duration and method | |
| CN105867341A (en) | Online equipment health state self-detection method and system for tobacco processing equipment | |
| Sand et al. | Towards an inline quick reaction system for actuator manufacturing using data mining | |
| CN110781206A (en) | Method for predicting whether electric energy meter in operation fails or not by learning meter-dismantling and returning failure characteristic rule | |
| CN117973846A (en) | Enterprise risk prediction method and system based on industrial chain | |
| CN113569374A (en) | Method and system for evaluating manufacturability of steel product | |
| CN117171145A (en) | Analysis processing method, equipment and storage medium for enterprise management system data | |
| CN115169426B (en) | Anomaly detection method and system based on similarity learning fusion model | |
| CN114048592A (en) | Finish rolling whole-flow distributed operation performance evaluation and non-optimal reason tracing method | |
| CN108985564B (en) | QFD (quad Flat No-lead) and binary semantic-based FMEA (failure mode and effects analysis) method | |
| Zhang | Portrait analysis of power transmission line for smart grid based on external data association fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190816 |
|
| RJ01 | Rejection of invention patent application after publication |