CN106296401A - A kind of Strong association rule method for digging understood for stock market's operation logic - Google Patents
A kind of Strong association rule method for digging understood for stock market's operation logic Download PDFInfo
- Publication number
- CN106296401A CN106296401A CN201611038982.9A CN201611038982A CN106296401A CN 106296401 A CN106296401 A CN 106296401A CN 201611038982 A CN201611038982 A CN 201611038982A CN 106296401 A CN106296401 A CN 106296401A
- Authority
- CN
- China
- Prior art keywords
- stock
- association rule
- rule
- strong association
- item
- 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
- 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
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Abstract
The invention discloses a kind of Strong association rule method for digging understood for stock market's operation logic.The main thought of method is: after the amount of increase and amount of decrease numerical quantization of all stocks, and these values regard the set of correlation rule middle term as, and the quantized value of the recent adjacent short time sequence of every stock then regards the transaction of affairs as；Then with the thought of association rule algorithm, all stock certificate datas are added up and rule discovery；Finally obtain and meet support threshold and the of a relatively high Strong association rule of confidence level and analyzed, thus the operation logic of stock market is understood and makes explanations.The inventive method is with the difference of traditional association rule and method: the item of traditional association rule is unordered set, and what we's rule requirement was ordered into.It is actual that the inventive method combines stock, can relatively profound understanding stock market operation logic, provide decision support for Stock-operation.
Description
Technical field
The present invention relates to stock certificate data digging technology field, especially relate to a kind of for stock market's operation logic understand strong
Association rule mining method.
Background technology
Stock market is always numerous scholars and the object of study of investor, the prediction of stock price tendency since setting up
It it is the problem of investment and security theoretical circles common concern.Due to various by domestic and international political economy environment and enterprise self etc.
The impact of factor, among stock price is always at constantly fluctuating, the influence mode of various factors is the most extremely complex, so stock
Price ups and downs are unordered, have the biggest random walk.
Stock price is conducted in-depth research by a large amount of scholars, and its prediction principle is: by history and current stock
Data are analyzed, and are predicted the price trend that stock is following, thus provide reference frame for stock invester.Conventional stock valency
Lattice Forecasting Methodology has: time series analysis method, neural net prediction method, regression analysis, time series smoothing techniques, trend are bent
Line model method, Random time sequence Forecasting Methodology, Markov Pre-measurement and discriminant analysis predicted method etc..Share price internal system is tied
The complexity of structure, the polytropy of external factor determine the arduousness of Stock Market Forecasting, existing analyzing and predicting method application effect
Unsatisfactory.
Although having various factors to result in the bad predictability of stock, but we still can be by certain methods to stock
Operation logic have individual basic insight.Stock market's operation logic described in the inventive method does not implies that the operating mechanism of stock market,
And the logic combined before and after referring to advance versus decline width, the most common chasing after is risen to kill and is fallen, i.e. rise sharply today, and tomorrow and then rises sharply
Probability is the biggest；Otherwise today slumps, the probability slumped tomorrow is the biggest.
Summary of the invention
The invention discloses a kind of Strong association rule method for digging understood for stock market's operation logic.The main think of of method
Think be: after the amount of increase and amount of decrease numerical quantization of all stocks, these values regard the set of correlation rule middle term as, and every stock
The quantized value of recent adjacent short time sequence then regards the transaction of affairs as；Then with the thought pair of association rule algorithm
All stock certificate datas carry out adding up and rule discovery；Finally obtain and meet support threshold and of a relatively high the closing by force of confidence level
Connection rule is also analyzed, thus understands the operation logic of stock market and make explanations.
The inventive method is with the difference of traditional association rule and method: the item of traditional association rule is unordered collection
Close, and what we's rule requirement was ordered into.The inventive method combines the reality of stock, can run relatively profound understanding stock market
Logic, provides decision support for Stock-operation.
Assume that stock list is S, S=[S1, S2,…,Si,…,Sm], m is the quantity of stock in stock pond, in China
The quantity of city's stock or the quantity of listed stock of the U.S..The inventive method learns common rule for all of stock.This
The step of bright method is as follows:
(1) the ups and downs amplitude of all stocks is quantified；
(2) adjacency to all stocks carries out orderly frequent item set statistics；
(3) stock market's Strong association rule is extracted based on support and confidence level sequence；
(4) stock market's operation logic understood and make explanations.
Wherein, the ups and downs amplitude to all stocks of step (1) quantifies, particularly as follows: for every stock, obtain certain
Amount of increase and amount of decrease data (such as on January 1st, 2005) since individual time point, then quantify ups and downs amplitude, i.e. to ups and downs amplitude
Carry out the operation rounded up, be transformed to the integer value that [-10,10] are interval；Every stock in last stock pond is all transformed to
Comprise the integer-valued array of advance versus decline.
Wherein, the adjacency to all stocks of step (2) carries out orderly frequent item set statistics, particularly as follows: with all stocks
The amount of increase and amount of decrease integer value of ticket is object of statistics, simultaneously on the basis of requiring to obtain adjacent quantization value in order, carries out k-item collection
Frequency statistics, k here takes 1 to 5.When carrying out k-item collection statistics, first one k dimension group of definition, the most one-dimensional is a length of
21, then travel through the amount of increase and amount of decrease integer value array of all stock, take k adjacent value every time, the common appearance that this k is worth
Number of times carries out cumulative statistics, and puts in the groove that k dimension group is corresponding.
Wherein, stock market's Strong association rule is extracted in sorting based on support and confidence level of step (3), particularly as follows: based on often
Group k item collection and the statistics of k-1 item collection, calculate the support under often organizing k item collection and confidence level, and k takes 2 to 5 here.Under k item collection
Support be the ratio that the secondary numerical value in the k each groove of dimension group accounts for all number of times sums；The molecule of confidence level is the system of k-item collection
Metering number, denominator is that k-item collection removes tail Xiang Houqi remainder in the k-1 item corresponding statistics number of concentration.To more than minimum support
Frequent item set sort from big to small according to confidence value, the rule corresponding to front t confidence value is Strong association rule.
For often organizing k item collection, the Strong association rule of acquisition is divided into former piece and consequent, after former piece refers to that tail item removed by k item collection
Residue other, consequent then refers to tail item, i.e. last.
Wherein, stock market's operation logic being understood and makes explanations, particularly as follows: obtain according to previous step of step (4)
The Strong association rule of number of packages before the various differences taken, learns these rules, selects significant rule, and when stock to be predicted is current
When recent transaction data matches the operation logic of stock market, i.e. before and after advance versus decline width, combine the upper rule extracted of arranging in pairs or groups
Time, prompting user pays close attention to, and informs the follow-up possible combination of history run logic and the probability occurred thereof.
Accompanying drawing explanation
Fig. 1 is the flow chart of the Strong association rule method for digging that the present invention understands for stock market's operation logic.
Fig. 2 is sub-fraction Strong association rule based on the inventive method output.Here rule is to carry out based on stock
The support obtained after data mining and the higher rule of confidence level.Owing to Strong association rule is more, the most only list little
A part.Rule can be used for helping to understand the operation logic of stock market.
Detailed description of the invention
Below in conjunction with the accompanying drawings and example, the present invention is described in detail.
Correlation rule is the implications of shape such as X → Y, X and Y is called former piece and the consequent of correlation rule.Wherein, association
, there is support and confidence level in rule XY.Correlation rule is defined as: assume I={I1, I2 ... .Im} is the set of item.Given one
Individual transaction data base D, the most each affairs t are the nonvoid subsets of I, i.e. each transaction is corresponding with a unique TID.
Correlation rule support (support) in D is the percentage ratio that in D, affairs comprise X, Y simultaneously, i.e. probability；Confidence level
(confidence) it is in the case of affairs have comprised X in D, comprises the percentage ratio of Y, i.e. conditional probability.If item collection meets
Minimum support threshold value, then this Xiang Jiwei frequent item set；If meeting minimum support threshold value and minimal confidence threshold, then recognize
It is interesting for correlation rule.These threshold values are manually set according to excavating needs.
After the amount of increase and amount of decrease numerical quantization of all stocks, these values regard the set of correlation rule middle term as；And every stock
The quantized value of ticket recent short time sequence is then considered as the transaction of affairs.It is actual that the inventive method combines stock, with
All stock certificate datas are added up and rule discovery by association rule algorithm thought, obtain and meet support threshold and confidence
Spend of a relatively high Strong association rule and analyzed, thus strengthening the understanding to stock market's operation logic.
The inventive method is with the difference of traditional association rule and method: the item of traditional association rule is unordered collection
Close, and what we's rule requirement was ordered into.
Assume that stock list is S, S=[S1, S2,…,Si,…,Sm], m is the quantity of stock in stock pond, in China
The quantity of city's stock or the quantity of listed stock of the U.S..Method learns common rule for all stocks.Stock market associates by force rule
Mining process then is specific as follows.
One, the ups and downs amplitude of stock is quantified.
For every stock, amount of increase and amount of decrease data (such as on January 1st, 2005) since obtaining certain time point, then to rising
Drop range value quantifies, the operation i.e. rounded up ups and downs amplitude, is transformed to the integer value that [-10,10] are interval；Finally
Every stock in stock pond is all transformed to comprise the integer-valued array of advance versus decline.
Two, stock frequent item set statistics.
2.1 carry out 1-item collection statistics.Owing to stock has amount of increase and amount of decrease to limit, the integer value after quantization has 21 kinds of situations, i.e. [-
10,10] interval integer, therefore arranges an array having 21 grooves, travels through the amount of increase and amount of decrease integer value array of all stock, according to
The integer-valued occurrence number of amount of increase and amount of decrease carries out cumulative statistics, and puts in the groove of correspondence.
2.2 carry out 2-item collection statistics.The two-dimensional array of one 21X21 is set, travels through the amount of increase and amount of decrease integer value of all stock
Array, takes adjacent two value every time, the common number of times occurred of the two value is carried out cumulative statistics, and puts into two-dimensional array
In corresponding groove.
2.3 carry out 3-item collection statistics.The three-dimensional array of one 21X21X21 is set, travels through the amount of increase and amount of decrease integer of all stock
Value array, takes adjacent three value every time, the common number of times occurred of these three value is carried out cumulative statistics, and puts into three dimensions
In the groove that group is corresponding.
2.4 carry out 4-item collection statistics.The four-dimensional array of one 21X21X21X21 is set, travels through advance versus decline width integer value
Array, takes adjacent four value every time, the common number of times occurred of these four values is carried out cumulative statistics, and puts into four-dimensional array
In corresponding groove.
2.5 carry out 5-item collection statistics.The five dimension groups of one 21X21X21X21X21 are set, travel through advance versus decline width integer
Value array, takes adjacent five value every time, the common number of times occurred of these five values is carried out cumulative statistics, and puts into five dimensions
In the groove that group is corresponding.
Owing to stock certificate data is limited, when typically proceeding to 5-item collection statistics, the frequency of occurrences of co-occurrence tuple is the least
, the higher-dimension array that statistics obtains is the most sparse, carries out necessity of higher item collection statistics the most again.
Three, stock market's Strong association rule is extracted.
Due to the particularity of stock application, be only concerned inside frequent item set last, i.e. next amount of increase and amount of decrease goes out the day of trade
Existing probability, is therefore different from traditional association rules mining algorithm, and the rear number of packages of the rule extracted here is all fixed as 1.
3.1 obtain 2-frequent item set and Strong association rule.2-item collection is calculated support, i.e. calculates two-dimensional array each
Secondary numerical value in groove accounts for the ratio of all number of times sums.Set a support threshold, filter out support less than support threshold
Groove, residual term collection is frequent item set.For each 2-item frequent item set, the residual term removing last is concentrated at 1-item
Obtain corresponding statistics number, the statistics number of 2-item collection divided by corresponding 1-item collection statistics number confidence level.To obtaining
The all confidence values obtained sort from big to small, and the rule corresponding to front t confidence value is Strong association rule.
Strong association rule the following table is some the 1-former pieces obtained.
3.2 obtain 3-frequent item set and Strong association rule.3-item collection is calculated support, i.e. calculates three-dimensional array each
Secondary numerical value in groove accounts for the ratio of all number of times sums.Set a support threshold, filter out support less than support threshold
Groove, residual term collection is frequent item set.For each 3-item frequent item set, the residual term removing last is concentrated at 2-item
Obtain corresponding statistics number, the statistics number of 3-item collection divided by corresponding 2-item collection statistics number confidence level.To obtaining
The all confidence values obtained sort from big to small, and the rule corresponding to front t confidence value is Strong association rule.
Strong association rule the following table is some the 2-former pieces obtained.
3.3 obtain 4-frequent item set and Strong association rule.4-item collection is calculated support, i.e. calculates four-dimensional array each
Secondary numerical value in groove accounts for the ratio of all number of times sums.Set a support threshold, filter out support less than support threshold
Groove, residual term collection is frequent item set.For each 4-item frequent item set, the residual term removing last is concentrated at 3-item
Obtain corresponding statistics number, the statistics number of 4-item collection divided by corresponding 3-item collection statistics number confidence level.To obtaining
The all confidence values obtained sort from big to small, and the rule corresponding to front t confidence value is Strong association rule.
Strong association rule the following table is some the 3-former pieces obtained.
3.4 obtain 5-frequent item set and Strong association rule.5-item collection is calculated support, i.e. calculates five dimension groups each
Secondary numerical value in groove accounts for the ratio of all number of times sums.Set a support threshold, filter out support less than support threshold
Groove, residual term collection is frequent item set.For each 5-item frequent item set, the residual term removing last is concentrated at 4-item
Obtain corresponding statistics number, the statistics number of 5-item collection divided by corresponding 4-item collection statistics number confidence level.To obtaining
The all confidence values obtained sort from big to small, and the rule corresponding to front t confidence value is Strong association rule.
Strong association rule the following table is some the 4-former pieces obtained.
Four, stock market's operation logic understands and explains.
According to the Strong association rule of number of packages before the various differences that previous step obtains, learn these rules, select significantly
Rule.Knowable to the Strong association rule of number of packages before any of the above difference, generally stock market is to become surely, i.e. next friendship
The amount of increase and amount of decrease integer value of Yi is all close to 0.When stock limit-up (amount of increase 10%), there is the biggest probability meeting next day of trade
Continue limit-up；When stock limit down (drop range-10%), next day of trade there is the biggest probability and may proceed to limit down；This also illustrates
Chasing after in stock market is risen and is killed that to fall be reasonable.Owing to Strong association rule is more, illustrate the most one by one.
Operation logic based on stock, i.e. stock Strong association rule, when the most recent transaction data of stock to be predicted
When matching the operation logic of stock market, i.e. before and after advance versus decline width combination can arrange in pairs or groups go up extraction regular time, at this moment point out use
Family is paid close attention to, and informs the follow-up possible combination of history run logic and the probability occurred thereof.
In sum, the invention discloses a kind of Strong association rule method for digging understood for stock market's operation logic.Side
All stock certificate datas are added up and rule discovery by method with the thought of association rule algorithm.It is different from traditional association rule and method
Item be unordered set, process require that what item was ordered into, and the consequent of rule only have an item.Method combines stock
Reality, can relatively profound understanding stock market operation logic, for Stock-operation provide decision support.
The inventive method is similarly applied to security class and has the data of time series feature, such as fund, futures etc..Cause
This, although disclosing the specific embodiments and the drawings of the present invention for the purpose of illustration, its object is to help to understand that the present invention's is interior
Hold and implement according to this, but it will be appreciated by those skilled in the art that: without departing from the present invention and the essence of appended claim
In god and scope, various replacements, to change and modifications be all impossible.Therefore, the present invention should not be limited to most preferred embodiment and
Accompanying drawing disclosure of that.Presently disclosed embodiment should be understood illustrative in all respects rather than it be claimed
The restriction of scope.
Claims (4)
1. the Strong association rule method for digging understood for stock market's operation logic, it is characterised in that described method includes as follows
Step:
(1) the ups and downs amplitude of all stocks is quantified；
(2) adjacency to all stocks carries out orderly frequent item set statistics；
(3) stock market's Strong association rule is extracted based on support and confidence level sequence；
(4) stock market's operation logic understood and make explanations.
The Strong association rule method for digging understood for stock market's operation logic the most according to claim 1, it is characterised in that
When all stocks are associated rule frequent episode statistics, the item of statistics is the quantized value of amount of increase and amount of decrease, and is by item
Carry out what Ordered Statistic adjacent for k-completed.
The Strong association rule method for digging understood for stock market's operation logic the most according to claim 1, it is characterised in that
During calculating correlation rule often organizes support and the confidence level of k item collection, it is only necessary to utilize k item collection and the statistics of k-1 item collection
Result carries out simple division arithmetic and can be quickly completed.
The Strong association rule method for digging understood for stock market's operation logic the most according to claim 1, it is characterised in that
In the extraction of Strong association rule, define that the consequent of rule has been only capable of an item, the most i.e. met next day of trade of stock
Prediction need, the acquisition process of Strong association rule can be made again to be greatly simplified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611038982.9A CN106296401A (en) | 2016-11-24 | 2016-11-24 | A kind of Strong association rule method for digging understood for stock market's operation logic |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611038982.9A CN106296401A (en) | 2016-11-24 | 2016-11-24 | A kind of Strong association rule method for digging understood for stock market's operation logic |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106296401A true CN106296401A (en) | 2017-01-04 |
Family
ID=57720970
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611038982.9A Pending CN106296401A (en) | 2016-11-24 | 2016-11-24 | A kind of Strong association rule method for digging understood for stock market's operation logic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106296401A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108090220A (en) * | 2017-12-29 | 2018-05-29 | 科大讯飞股份有限公司 | Point of interest search sort method and system |
| CN109508386A (en) * | 2018-11-07 | 2019-03-22 | 福建工程学院 | A kind of relevance metric method of stock information press center word and related stock |
| CN109697619A (en) * | 2017-10-20 | 2019-04-30 | 中移(苏州)软件技术有限公司 | A kind of data analysing method and device |
| CN111046090A (en) * | 2020-03-10 | 2020-04-21 | 深圳开源互联网安全技术有限公司 | Vehicle data mining method and system based on vehicle-mounted self-organizing network |
-
2016
- 2016-11-24 CN CN201611038982.9A patent/CN106296401A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109697619A (en) * | 2017-10-20 | 2019-04-30 | 中移(苏州)软件技术有限公司 | A kind of data analysing method and device |
| CN108090220A (en) * | 2017-12-29 | 2018-05-29 | 科大讯飞股份有限公司 | Point of interest search sort method and system |
| CN108090220B (en) * | 2017-12-29 | 2021-05-04 | 科大讯飞股份有限公司 | Method and system for searching and sequencing points of interest |
| CN109508386A (en) * | 2018-11-07 | 2019-03-22 | 福建工程学院 | A kind of relevance metric method of stock information press center word and related stock |
| CN111046090A (en) * | 2020-03-10 | 2020-04-21 | 深圳开源互联网安全技术有限公司 | Vehicle data mining method and system based on vehicle-mounted self-organizing network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Paithankar et al. | Open pit mine production schedule optimization using a hybrid of maximum-flow and genetic algorithms | |
| CN106296401A (en) | A kind of Strong association rule method for digging understood for stock market's operation logic | |
| CN107291895B (en) | Quick hierarchical document query method | |
| Francés et al. | The cryptocurrency market: A network analysis | |
| Guo | Oil price forecast using deep learning and ARIMA | |
| CN110826618A (en) | Personal credit risk assessment method based on random forest | |
| CN113052225A (en) | Alarm convergence method and device based on clustering algorithm and time sequence association rule | |
| Amayuelas et al. | Neural methods for logical reasoning over knowledge graphs | |
| CN111967971A (en) | Bank client data processing method and device | |
| CN108038211A (en) | A kind of unsupervised relation data method for detecting abnormality based on context | |
| CN106780018A (en) | A kind of stock certificate data mining model reappeared based on history and system | |
| Liu | Incomplete big data imputation mining algorithm based on BP neural network | |
| Li et al. | Chart GCN: Learning chart information with a graph convolutional network for stock movement prediction | |
| Gupta et al. | A classification method to classify high dimensional data | |
| Malik et al. | A comprehensive approach towards data preprocessing techniques & association rules | |
| Garciarena et al. | Evolving imputation strategies for missing data in classification problems with TPOT | |
| CN105678430A (en) | Improved user recommendation method based on neighbor project slope one algorithm | |
| Wang | Predicting the rise and fall of Shanghai composite index based on artificial intelligence | |
| CN113378842A (en) | Recommendation method based on segmented image feature extraction | |
| Yang | RETRACTED ARTICLE: Financial Information Extraction Using the Improved Hidden Markov Model and Deep Learning | |
| Iyer et al. | Comparison of logistic regression and artificial neural network based bankruptcy prediction models | |
| CN111984842A (en) | Bank client data processing method and device | |
| CN109766332A (en) | Data processing empty value method, apparatus and terminal device | |
| Khatun et al. | An efficient genetic algorithm for discovering diverse-frequent patterns | |
| Moodi et al. | Evaluation of feature selection performance for identification of best effective technical indicators on stock market price prediction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170104 |