Font Size: a A A

The Application Research Of Time Series Data Mining On Securitties Market Prediction

Posted on:2006-01-15Degree:MasterType:Thesis
Country:ChinaCandidate:H GuFull Text:PDF
GTID:2168360155453009Subject:Software engineering
Abstract/Summary:PDF Full Text Request
With the arrival of information age, the prevalence of storing explosibility increasing and network of datum make people fall in ocean of the data and information. Wide application of the rapid development of the technology of the database and data base management system,make the data that people accumulated get more and more.A lot of important information is being hidden behind one's back in datum that increase sharply, people hope to analysis to it more high-level,in order to utilize these data better.But ,at present,the system of the database can only access the existing data, but unable to find the hidden knowledge behind the data. cause the phenomenon that " the data explode but knowledge is poor ". It is a very important subject how to find and draw valuable knowledge in the huge data. Knowledge Discoveryin Databases and Data Mining are important method to find ,comprehend and utilize the knowledge from the datebases effectively. Date mining is a lot of result that discipline combine together of the technology of databases, Artificial Intelligence , machine study,artificial neural networ,pattern-recognition,statistical analysis, confuse logic. Time Series Data is a focuses among them because their extensive application at present and extremely high commercial value. We analyse the characteristic of the securities market, find the securities market has unitarity of data (a large number of date does not need processed with special-method ), characteristic of analysing the means variety and disguise, accord with data excavate to have a large number of sufficient data requirement, need characteristic , knowledge of decision, therefore point out that date mining is feasibility and urgency applied to security analysis of the data, and then analyse through the characteristic to the securities , have proposed the model of date mining suitable for the securities data. The theory with importance of prediction question of the Time array in system on the securities market and actual meaning . any reason of people's expects in essence behaviors are all to predict future according to history.In fact , the future is not include in history. In the stock market .It is expected behavior to make a deal situation similar historical stability of orbit determine to oneself their or not. Regard as friendship when condition and its historical dependence of orbit strong,namely the former at having stronger regularity reliance in the latter, the reason based on this predicts the behavior is rational,effectual; On the contrary , have no foundation and not reliable one. Obtain time array is through analyse and win to gather to classify from the database. Obtain time array can predict that to it after the data. Prediction and analyses of time array is include Long term trend,Cyclical component, Seasonal component, Irregular component. We,through the analysis of the array to time can finish forecasting,modeling,characterization.The purpose of forecastingis to predict the systematic gradual progress more accurately. The purpose of modeling mould can catch the description of th e long-term behavior characteristic of the system to provide; The purpose of characterization.is drawn is to confirm the ba sic attribute of the array on terms that there is no priori k nowledge. A large number of data in the stock market are a quantity that change in succession. It is a not easy thing to deal with so huge ,increased data .we would attempt treatment succession to dispersed , and carry on mode analysis to the characteristic amount after being dispersed. Through analysis and treatment of data of time array realize to stock market future prediction of tendency correctly. At present ,the treatment to time series of the stock market is concentrated on two problems mainly,the first is the search for the similar array, another one is knowledge is found of the time series. Before searching for , click the description with ups and downs characteristic in the stock market to draw first.And match one character bunch with characteristic. it is a sub array that will carry on search for. carry on the search for the sub array in showing the overall array of all stock data, carry on the mode to analyse to the result after searching for, it is similar in order to expect to be able to find historical data. At present,there have a lot of method of match character bunches . DNA array as a rray language that include 4 space-time control information.It be made from 4 letter as A,G,T,C, Carry onchange the financial time array into character array,we must consider price go up and down a important one number value reflect most for security market, there are differences to hoist or lower of the size,can't treat as the same. We introduce 4 kinds of character R , r , d , D, Among them, R representative the mode of rise soon; r Representatives the mode of rise generally; d Representatives the mode of lower generally D Representatives the mode of lower quickly .therefore we turn a time array of the financial price into a character array that express with R ,r,d and D.We can find out that identification character R , r , d , D have clear physics intension. In this thesis, we carry on some improvement to existing DNA time array,change DNA array into the method that can representative computer data mostly--The binary system.The information of the whole stock market expressed uses 0 and 1, so it can reduce the complexity matched and searching for. We adopt regular expression while the character bunch searches for.Regular expression is describing a kind of mode to matched character bunch. Can used for checking if the character bunch .include some sub bunches,or replace some sun bunch. Can be interpreted as too: regular expression is a formula that match character bunch with a certain mode. In the simplest condition, regular expression seem to be one finding out bunch ordinary. Through the experiment,we search for the sub character bunch with a certain characteristic quantity, After carrying on comparative analysis to a large number of experimental datas,we...
Keywords/Search Tags:Data Mining, Time Series, Regular Expression, Securities Analysis, Character Bunch Matched, Similarity Search
PDF Full Text Request
Related items