فرکسیون : الگوریتم تلفیقی جدید پیش بینی روند شاخص صنایع بازار بورس و اوراق بهادار تهران

در اﯾﻦ ﻣﻘﺎﻟﻪ ﺑﻪ ﻣﻨﻈﻮر ﺑﻬﺒﻮد دﻗﺖ ﭘﯿﺶﺑﯿﻨﯽ روﻧﺪ ﺷﺎﺧﺺ ﺻﻨﺎﯾﻊ ﻣﺨﺘﻠﻒ در ﺑﺎزار ﺑﻮرس، اﻟﮕﻮرﯾﺘﻢ ﺗﺮﮐﯿﺒﯽ ﺟﺪﯾﺪ ﻓﺮﮐﺴﯿﻮن، ﮐﻪ ﺗﻠﻔﯿﻘﯽ از دو اﻟﮕﻮرﯾﺘﻢ دروﻧﯿﺎﺑﯽ ﻓﺮﮐﺘﺎل و رﮔﺮﺳﯿﻮن ﻣﺎﺷﯿﻦ ﺑﺮدار ﭘﺸﺘﯿﺒﺎن اﺳﺖ، ﭘﯿﺸﻨﻬﺎد ﻣﯽﺷﻮد. ﺑﺮای رﺳﯿﺪن ﺑﻪ اﯾﻦ ﻣﻬﻢ، ﺑﺎ ﻣﻌﺮﻓﯽ اﻟﮕﻮرﯾﺘﻢ دروﻧﯿﺎﺑﯽ ﻓﺮﮐﺘﺎل اﺻﻼح ﺷﺪه و ﭘﯿﺎدهﺳﺎزی آن ﺑﺮ روی دادهﻫﺎی اوﻟﯿﻪ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺷﺎﺧﺺ ﯾﮏ ﺻﻨﻌﺖ ﻓﺮﮐﺘﺎﻟﯽ، ﺑﻪ ﺗﻮﻟﯿﺪ دادهﻫﺎی ﻓﺮﮐﺘﺎﻟﯽ ﺟﺪﯾﺪ ﻣﺮﺑﻮط ﺑﻪ آن ﺻﻨﻌﺖ ﻣﯽﭘﺮدازﯾﻢ. در اداﻣﻪ، ﺑﺎ ﻓﺮاﺧﻮاﻧﯽ اﻟﮕﻮرﯾﺘﻢ رﮔﺮ ﺳﯿﻮن ﺑﺮدار ﭘ ﺸﺘﯿﺒﺎن ﺑﺮ روی دادهﻫﺎی ﻓﺮﮐﺘﺎﻟﯽ ﺟﺪﯾﺪ، ﺑﻪ ﭘﯿﺶﺑﯿﻨﯽ روﻧﺪ ﺷﺎﺧﺺ آن ﺻﻨﻌﺖ ﺧﻮاﻫﯿﻢ ﭘﺮداﺧﺖ. ﻧﺘﺎﯾﺞ ﺣﺎﺻﻞ از ﭘﯿﺎدهﺳﺎزی اﻟﮕﻮرﯾﺘﻢ ﺗﺮﮐﯿﺒﯽ ﻓﺮﮐﺴﯿﻮن ﺑﺮ روی دادهﻫﺎی ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺷﺎﺧﺺ ﺻﻨﺎﯾﻊ ﻣﺨﺘﻠﻒ در ﺑﺎزار ﺑﻮرس و اوراق ﺑﻬﺎدار اﯾﺮان و ﻣﻘﺎﯾ ﺴﮥ ﻋﻤﻠﮑﺮد آن ﺑﺎ ﻋﻤﻠﮑﺮد دو روش ﻣﺮ ﺳﻮم ﯾﻌﻨﯽ ﺷﺒﮑﻪ ﻋ ﺼﺒﯽ ﻣ ﺼﻨﻮﻋﯽ و رﮔﺮ ﺳﯿﻮن ﻣﺎ ﺷﯿﻦ ﺑﺮدار ﭘ ﺸﺘﯿﺒﺎن، ﺣﺎﮐﯽ از ﺑﺮﺗﺮی دﻗﺖ ﭘﯿﺶﺑﯿﻨﯽ اﻟﮕﻮرﯾﺘﻢ ﭘﯿﺸﻨﻬﺎدی اﺳﺖ.

fracsion: new hybrid algorithm for predicting industrialindex trends in tehran stock exchange

stock market research shows that this market is a complex nonlinear dynamic system. therefore, investing in this market is always associatedwith various risks, and therefore predicting the market trend is veryimportant. with the rapid development of artificial intelligence theory andmachine learning methods in recent years, some intelligent methods suchas artificial neural network (ann) and machine support vector (svm)have been widely used to model and predict financial time series.the svm method is a supervised machine learning algorithm used tosolve classification and regression problems. initially, the svm methodwas used only to solve classification problems, but later this algorithm wasdeveloped to work with regression or data estimation problems and wascalled support vector regression (svr). on the other hand, in theeconomic and social sectors, owing to a large number of non-stationaryand irregular data, fractal model, especially, fractal interpolation model isoften used to fit the missing data and to predict the short-term trends.in this paper, to improve the accuracy of predicting the trend of variousindustries in the stock market, a new hybrid fracsion algorithm, which isa combination of two fractal interpolation algorithms and support vectorregression, is proposed.