Fo recasting Stock M arket Q uo tation s via Fuzzy N eu ral N etw o rk Based on T 2S M odel

Σχετικά έγγραφα
D esign and Imp lem en tation of Parallel Genetic A lgo rithm

T he Op tim al L PM Po rtfo lio M odel of H arlow s and Its So lving M ethod

GPS, 0. 5 kg ( In tegrated Fertility Index, IF I) 1. 1 SPSS 10. IF I =

A R, H ilbert2h uang T ran sfo rm and A R M odel

Con struction and D emon stration of an Index System of Know ledge Amoun t of Po sition

E stab lish ing Syn thesis Evaluation Index

M athem aticalm odel and A lgo rithm of In telligen t T est Paper

A pp lication Study on R econ struction of Chao tic T im e Series and P rediction of Shanghai Stock Index

Bank A ssetgl iab ility Sheet w ith Em bedded Op tion s

copula, 5 3 Copula Κ L = lim System s Engineering M ay., 2006 : (2006) ,,, copula Ξ A rch im edean copula (Joe,

A Study of the O rigin of Comp lex ity in the Science of Comp lex ity

A multipath QoS routing algorithm based on Ant Net

China Academic Journal Electronic Publishing House. All rights reserved. O ct., 2005

D ecision2m ak ing M odel Inco rpo rating R isk Behavio r under P ro ject R isk M anagem en t

Im pact of capac ity a lloca tion on bullwh ip effect in supply cha in

Application of a novel immune network learn ing algorithm to fault diagnosis

O verlay A lgo rithm of H ierarch ical M ap s

Error ana lysis of P2wave non2hyperbolic m oveout veloc ity in layered media

V o l122, N o13 M ay, 2003 PRO GR ESS IN GEO GRA PH Y : (2003) , : TU 984. , (Eco logy fo r evil) [1 ] (R ich Boyer), 2.

The Optim ization A lgor ithm s for Solv ing Resource-con stra ined Project Schedul ing Problem: A Rev iew

On- l ine com puter detecting system of p ipel ine leak and its algor ithm

Q π (/) ^ ^ ^ Η φ. <f) c>o. ^ ο. ö ê ω Q. Ο. o 'c. _o _) o U 03. ,,, ω ^ ^ -g'^ ο 0) f ο. Ε. ιη ο Φ. ο 0) κ. ο 03.,Ο. g 2< οο"" ο φ.

U (x, y ) = : K (x i- x k) K (x i- x k, y j- y l), 2. 1

1 (forward modeling) 2 (data-driven modeling) e- Quest EnergyPlus DeST 1.1. {X t } ARMA. S.Sp. Pappas [4]

Buried Markov Model Pairwise

F ig. 1 Flow chart of comprehen sive design in the research of com patibil ity. E2m ail: ac. cn

A Knowledge M odel for D esign of Population Nutr ien t Index D ynam ics in W heat

: Monte Carlo EM 313, Louis (1982) EM, EM Newton-Raphson, /. EM, 2 Monte Carlo EM Newton-Raphson, Monte Carlo EM, Monte Carlo EM, /. 3, Monte Carlo EM

Jou rnal of M athem atical Study

26 3 V o l. 26 N o A cta Eco logiae A n im alis Dom astici M ay ,

The Research on Sampling Estimation of Seasonal Index Based on Stratified Random Sampling

我国股票市场指数及指数证券投资组合 关键词 中图分类号 文献标识码 文章编号 0 引 言

F ingerpr in ts of X in shu Ora l L iquid by HPCE

2016 IEEE/ACM International Conference on Mobile Software Engineering and Systems

W eb. W eb Information Extraction Based on Tree Structure. REN Zhong- sheng 1, XUE Y ong- sheng 2

ED SS (Environm en tal D ecision Suppo rt

S ingula r C onfigura tion Ana lys is a nd C oo rd ina te C ontro l of Robo t

TR IBOLO GY N ov, 2004 , P I2. , 120 M Pa, P I. 6 mm 7 mm 30 mm [7, 8 ] mm,. JSM 25600LV 1. 1 (EDXA ).

Matrices and vectors. Matrix and vector. a 11 a 12 a 1n a 21 a 22 a 2n A = b 1 b 2. b m. R m n, b = = ( a ij. a m1 a m2 a mn. def

Estimation of grain boundary segregation enthalpy and its role in stable nanocrystalline alloy design

( , ,

M p f(p, q) = (p + q) O(1)

The Changes of Card iova scular Respon se to O rthosta tic Stress Caused by Hypovolem ia Inuced by W e ightlessness: A Sim ula tion Study

1. 2 , N RC (1998) d m g 1

Durbin-Levinson recursive method

On Channel-adaptive Error Con trol Techn ique V ideo Comm un ication

hm 2,, , hm 2, A CTA GEO GRA PH ICA S IN ICA

(H ipp op hae rham noid es L. ) , ; SHB 2g , 1. 0, 2. 0, 3. 0, 4. 0, 5. 0 ml mm. : Y = X - 0.

VSC STEADY2STATE MOD EL AND ITS NONL INEAR CONTROL OF VSC2HVDC SYSTEM VSC (1. , ; 2. , )

ADVAN CE IN EA R TH SC IEN CES : (2000)

6.3 Forecasting ARMA processes

( P- V EP, ER G K8,A g- A gc1. V EP s. : oh z 30H z; : 0. 24, 0. 48, 0. 96, 1. 85, %

Nov Journal of Zhengzhou University Engineering Science Vol. 36 No FCM. A doi /j. issn

6 cm, 1. 2 IAA, NAA, 2. 1

P rogresses in H azardou s M aterials L ogistics R esearch

Ch inese Journal of M anagem ent

) , EPR I [ 5, 6 ], [ 1 ], ; ; ; ; ; ; ] Tsinghua Tongfang Optical Disc Co., Ltd. All rights reserved.

Retrieval of Seismic Data Recorded on Open-reel-type Magnetic Tapes (MT) by Using Existing Devices

Schedulability Analysis Algorithm for Timing Constraint Workflow Models

Παλεπηζηήκην Πεηξαηώο Τκήκα Πιεξνθνξηθήο Πξόγξακκα Μεηαπηπρηαθώλ Σπνπδώλ «Πξνεγκέλα Σπζηήκαηα Πιεξνθνξηθήο»

EL ECTR IC MACH IN ES AND CON TROL. System s vulnerability assessment of a ircraft guarantee system based on improved FPN

Appendix to On the stability of a compressible axisymmetric rotating flow in a pipe. By Z. Rusak & J. H. Lee

Simplex Crossover for Real-coded Genetic Algolithms

M in ing Recursive Function s Ba sed on Gene Expression Programm ing

1999, 17 (1): J ourna l of W uhan B otan ica l Resea rch ( ) ( ) 2, 3. (Celosia cristata L. ),

March 14, ( ) March 14, / 52

, 4, 6, 8 m in; 30%, 50%, 70%, 90% 100% 15, 30, 45, 60, s V c

17 1 V o l. 17 N o CH IN ESE JOU RNAL O F COM PU TA T IONAL M ECHAN ICS February 2000 : A

MATH 38061/MATH48061/MATH68061: MULTIVARIATE STATISTICS Solutions to Problems on Matrix Algebra


ο ο 3 α. 3"* > ω > d καΐ 'Ενορία όλις ή Χώρί ^ 3 < KN < ^ < 13 > ο_ Μ ^~~ > > > > > Ο to X Η > ο_ ο Ο,2 Σχέδι Γλεγμα Ο Σ Ο Ζ < o w *< Χ χ Χ Χ < < < Ο

Approximation of distance between locations on earth given by latitude and longitude

3607 Ν. 7.28/88. E.E., Παρ. I, Αρ. 2371,

Αλγοριθμική ασυμπτωτική ανάλυση πεπερασμένης αργής πολλαπλότητας: O ελκυστής Rössler

Matrices and Determinants

, ( , (P anax qu inquef olius) ) ; L C2V P series U H PE), M Pa R E252. C18 (5 Λm, 250 mm 4. 6 mm ) ( )

, E, PRO GR ESS IN GEO GRA PH Y. V o l122, N o15 Sep t1, 2003 : (2003) , 263, (0 1m ) P93511; P42616

(α) Στη στήλη «Θέσεις 1993» ο αριθμός «36» αντικαθίσταται. (β) Στη στήλη των επεξηγήσεων αναγράφεται η ακόλουθη

o l. 26 N o Jou rnal of N an jing In stitu te of M eteo ro logy Feb. 2003

derivation of the Laplacian from rectangular to spherical coordinates

Example Sheet 3 Solutions

ITU-R P (2012/02)

[4] 1.2 [5] Bayesian Approach min-max min-max [6] UCB(Upper Confidence Bound ) UCT [7] [1] ( ) Amazons[8] Lines of Action(LOA)[4] Winands [4] 1

ΑΡΙΣΤΟΤΕΛΕΙΟ ΠΑΝΕΠΙΣΤΗΜΙΟ ΘΕΣΣΑΛΟΝΙΚΗΣ

Commutative Monoids in Intuitionistic Fuzzy Sets

X g 1990 g PSRB

B id irectional Ind irect Coupled F in ite Elem en tm ethod for Estimating Am pac ity of Power Cable

Ax = b. 7x = 21. x = 21 7 = 3.

Numerical Analysis FMN011

(T rip tery g ium w ilf ord ii Hook) ,Beroza [6 ] 4 W ilfo rine W ilfo rdine W ilfo rgine W il2. Euon ine 1. 0% 1980, 1. 1 ; 1. 0%, ; 0.

Knowledge Induction Ba sed on Genera liza tion of M ulti2benchmark A ttr ibute

«Συμπεριφορά μαθητών δευτεροβάθμιας εκπαίδευσης ως προς την κατανάλωση τροφίμων στο σχολείο»

Sarò signor io sol. α α. œ œ. œ œ œ œ µ œ œ. > Bass 2. Domenico Micheli. Canzon, ottava stanza. Soprano 1. Soprano 2. Alto 1

V isualization of the functional or ien tation column s in the cat v isual cortex by in vivo optical imag ing based on in tr in sic signals

Vol. 31,No JOURNAL OF CHINA UNIVERSITY OF SCIENCE AND TECHNOLOGY Feb

Exercises 10. Find a fundamental matrix of the given system of equations. Also find the fundamental matrix Φ(t) satisfying Φ(0) = I. 1.


Analysis and design of printed IFA for W HAN

Arbitrage Analysis of Futures Market with Frictions

!)'*! ' !"#$$%$$" &'( )'&* +((!"#$$%* +((* ++(, % 6 +7" "%!$ " $%&

Transcript:

2001 2 2 : 100026788 (2001) 0220066207 T 2S,, (, 400044) : T 2S,,, ( ),,. : ; ; α Fo recasting Stock M arket Q uo tation s via Fuzzy N eu ral N etw o rk Based on T 2S M odel CH EN X ing, M EN G W ei2dong, YAN T ai2hua (Schoo l of Bu siness & A dm in istration, Chongqing U n iversity, Chongqing 400044) Abstract T h is paper p resents a m ethod fo r stock m arket modeling and fo recasting via fuzzy neural netw o rk based on T 2S model, in w h ich the imp roved genetic algo rithm is used to train the connection w eigh ts of the fuzzy neural netw o rk, the algo rithm s of mom entum and self- adap tive learning rate are used to learn m em bersh ip param eterṡ It has been show n by the modeling and fo recasting results about Shanghai stock m arket p rice index and X iax in electron p rice ( a com pany stock p rice) that the m ethod has reinfo rcem en t learn ing p roperties, m app ing capab ilities, reflecting ab ility. W ith respect to modeling and fo recasting o r relative decision of stock m arket o r som e o ther sim ilar nonlinear econom ic system, the m ethod is available. Keywords stock m arket; fuzzy neural netw o rk; fo recasting 1,,,,,,,,,,,,,,,, (A rtificialn euraln etw o rk s, ANN ) [1 5, ] ANN,, ANN, α : 1999206202

2 T 2S 67,,,,,,,,, (Fuzzy N eural N etw o rk s, FNN ), [3, 6 ], FNN,,,,,,,, (Genetic A lgo rithm s, GA ),, [7 9 ] T 2S,,, ( ), 2 T-S, :, NB, PB ;,. T akagi Sugeno, T 2S. X x i T (x i) = {A 1 i,a 2 i,, A m i },, 2,, n. A s i i (s, 2,,m i) x i si, x i u s i A i (x i) (, 2,, n; s, 2,,m i) Y, T 2S R j: IF x 1 is A s 1j 1, x 2 is A s 2j 2,, x n is A s nj n TH EN y j = W j0 + W j1 x 1 + W j2 x 2 + + W jn x n j = 1, 2,,m, m Φ 7 n m i, X, a j = u s 1j A 1 (x 1) u s 2j A 2 (x 2) A n (x n), u s nj Y = a jy j a j = a λ jy y, a λ j = a j, a j 1,,, f (q) (q- 1) (q- 1) (q- 1) j (x 1, x 2,, x n q- 1 ; w j1 (q), w q j2,, w q jn q- 1 ) x (q) j = g (q) j, j g (q) j f (q) (f (q) j ) q j, x i, f (1) i = x (0) = x i, x (1) i = g (1) i = f (1) i, i= 0, 1,, n; 0 x 0= 1, W j0, j = 1, 2,,m N 1= n+ 1 1 - i, i = u s i A i,, 2,, n; s, 2,,m in,m i si u s i u s i (1) (x, f (2) i - c ) 2 =, x (2) = u s i i = g (2) = e f (2) e - (x i - c isi ) 2 Ρ 2,, 2,, n; s, 2,, m i C Ρ, Ρ (2) =

68 2001 2 1 T - S N 2 = 6 n m i 2,, f (3) j = m in{x (2) 1s 1j, x (2) 2s 2j,, x (2) ns nj }= m in u s 1j 1, u s 2j 2,, u s nj n }, s1j {1, 2,,m 1}, s2j {1, 2,,m 2},, snj {1, 2,, m n} x (3) j = a j= g (3) j = f (3) j, j = 1, 2,,m, m = 7 n m i N 3 = m,, 3 f (4) j = x (3) j x (3) i = a j a i, x (4) j = a λ j = g (4) f (4) j, a λ j= a j a, 2,,m. N 4 = m m,,, y j= W j0+ W j1 x 1+ W j2 x 2+ + W jn x n= 6 n W j lx l, j = 1, 2,,m. l= 0, Y = x (5) = g (5) = f (5) = a jy j, Y. j = a j= j = 1a λ jy j, a λ j = a j a j,, 3 3, W j i (j = 1, 2,, m ; i= 0, 1,, n) 1 c Ρ (, 2,, n; si = 1, 2,,m i).

2 T 2S 69 E = 1 2 (Yϖ - Y ) 2, ϖ Y Y 3. 1 c Ρ c Ρ, (1) D (k) = c (k + 1) = c (k) + Α(k) [ (1 - Γ)D (k) + ΓD (k - 1) ] (1) Ρ (k + 1) = Ρ (k) + Α(k) [ (1 - Γ)D (k) + ΓD (k - 1) ] (2) - 5E 5c (k), (2) D (k) = - 5E 5Ρ (k), Α (k) = 2 Κ Α(k+ 1), Κ= sign [D (k)d (k - 1) ],, 2,, n, s, 2,,m iα,,, ;,, Γ, 0Φ Γ< 1.,, 3. 2 W j i W j i M ich igan John H. Ho lland, A dapation in N ature and [8 A tificial System ] (, ),,,, Go ldberg SGA (Simp le GA, SGA ), :, ;, ;, ;, SGA, 2 1),,,, (W 10, W 11,,W 1n, W 20, W 21,, W 2n,, W m 0, W m 1,, W m n), 2) f,, f = 1 6 j E 2 j 3) :, e - gχg 4) :, P s : P s = f i 6 f i, f i i. 5) : P c ( ), P c,, P c, 2

P c = k c (f m ax - f c) 2 g(f m ax - f θ ) 2, f c Ε f θ k c f c < f θ k c 1, f c, f m ax f θ, f m ax- f θ,, 6) :, P m P m,,,, P m, km 70 2001 2 P m = km (f m ax - f m ) 2 g(f m ax - f θ ) 2, f m Ε f θ km f m < f θ 1, km 4 ( ),,,, : ( g ), :,, t,,,,,, M A CD WM S KDJ R S I B IA S PSY A R OBV ADR,,,,, 5 ( 5, (), 5, B IA S R S I A R K OBV : 1) x 1, 10 (B IA S (10) ) x 1= B IA S (10) = 2) x 2, 5 (M A (5) ) MA (5), B IA S30 x 2= B IA S = 6 n+ 13 MA (5) (30) = 3) x 3, 14 R S I(14) x 3= R S I(14) = i= n C 10- M A (10), C M A (10) 10 30 (M A (30) ), M A (5) B IA S (30), M A (5) - M A (30) M A (30) (C i - C i+ 1), C i+ 1- C i< 0; C i i 4) x 4, 26 (A R (26) ) x 4 = A R (26) = A A + B =, A = 6 n+ 13 (C i+ 1 - C i), C i+ 1- C i> 0; B 6 n+ 25 (H i - O i) 6 n+ 25 (O i - L i) i= n H i,o i,l i i ; 6 n+ 25 (H i - O i) 26, 6 n+ 25 (O i - L i) 26 5) x 5, K (K (9) ) x 5= K (today) = 2 3 K (yestaday) + 1 3 R SV (today) R SV (row

2 T 2S 71 stochastic value,,, WM S% ) 9 R SV, R SV : 9 R SV = C 9- L 9 H 9- L 9 100 C 9, H 9, L 9 9 6) x 6, 6 OBV (on balance vo lum e, ),OBV (6) x 6 = 6 n+ 5 2 ci - h i - li i= n h i - li li i,, v i : Y = M A (5) : ( (x 1, x 2, x 3, x 4, x 5, x 6) gy ), ( (B IA S (10), B IA S MA (5) (30), R S I(14), A R (26), K (9), OBV (6) ) gm A (5) ) v i ci, h i,, 5 c Ρ Γ= 0. 75 W j i, k c= 0. 15, km = 0. 08, N 80, 1997 10 31 1998 8 21 200, 1998 8 28 1998 10 30 50,, 2. 02, 78, 3. 85, 70 3 (, ), 1997 10 31 1998 8 21 200, 1998 8 28 1998 11 13 60,, 3. 15%, 75%, 5. 87%, 68% 4 (, ) 3 4,,,,, (,, ), ( ),,,,

72 2001 2,,, 1998,,, 9000 3,,,,,, 9000,, 5,,,,,,,,,,,,,, : [1 ] R efenes A N. et al. Stock perfo rm ance modeling using neural netw o rk s: a comparative w ith R egression M odels[j ]. N eural N etw o rk, 1994, 7 (2): 375 388. [2 ] L ee K C et al. Integration of hum an know ledge and m ach ine know ledge by using fuzzy po st adjustm ent: its perfo rm ance in stock m arket tim ing p rediction[j ]. Expert System, 1995, 12 (4): [3 ],. [M ]. :, 1993. [4 ]. [M ]. :, 1995. [ 5 ]. [J ]. ( ), 1998 (8): 25 30. [6 ]. T 2S [J ]. ( ), 1997 (3): 76 80. [7 ] Ho lland J. A dap tation in N atural and A rtificial System [M ]. A nn A rbo r : U niveṙ of M ich igan P ress, 1975. [8 ] Go ldberg D. Genetic A lgo rithm s in Search, Op tim ization and M ach ine L earning [ M ]. M assacheusettes: A ddision- W esley Pub., 1989 [9 ] W h itley D et al. Genetic algo rithm s and neural netw o rk s: op tim ization connections and connectivity [J ]. Parallel Computing, 1990, 14: 347 361 [10 ] T akagi T, Sugeno M. Fuzzy Identification of System s and its A pp lication to M odeling and Contro l [J ]. IEEE T ransaction on System s, M an, and Cybernetics, 1985, 15 (1): 116 132 [11 ]. [J ]., 1998 (9): 104 109. [12 ]. [J ]., 1999, 19 (2): 65 69. [13 ]. [D ]., 1999.