Study Of Time Series Data Based On Rough Set Theory

Rough set is a new valid mathematical theory developed in recent years, which has the ability to deal with imprecise, uncertain, and vague information. It can abtain valid and potentially useful knowledge in data. Applying rough set theory in data mining field can largely improve the analyzing and learning ability for incomplete data of large database, which has extensive applied prospect and value. Based on rough set theory the problems of time series processing have been mainly studied in this thesis.Research has been done to analyze time series using rough set method. It mainly contains the method for translating TIS to IS and translating RTTIS to TIS. Some methods of mining time series with rough set are discussed. In process of translating TIS to IS,traced time segments problem has already been an important problem of data mining from a time series with Rough Set. Further more a method for transforming TIS to IS based on time granularity is proposed.Attribute reduction is one of the key problems for the knowledge acquisition. In practicality time series database are dynamic,so study on incremental algorithms for attribute reduction is throughly important. Incremental algorithms for attribute reduction based on discernability matrix are proposed,by which attribute reduction of new decision table can be obtained quickly when records are added or deleted to primary decision table.Finally,through simulation,we validate the effect and accuracy of the algorithm of incremental reduct for dynamic database.