TY  - JOUR
T1  - Forecasting Stock Index Data Using Hybrid Models
AU - Vasimalla, Kumar AU - Narasimham, C. AU - , Deekshitha 
JO  - Journal of Engineering and Applied Sciences
VL  - 14
IS  - 8
SP  - 2752
EP  - 2763
PY  - 2019
DA  - 2001/08/19
SN  - 1816-949x
DO  - jeasci.2019.2752.2763
UR  - https://makhillpublications.co/view-article.php?doi=jeasci.2019.2752.2763
KW  - Hybrid model
KW  -Laplacian score
KW  -multi cluster
KW  -co-relation
KW  -ANFIS
KW  -Genetic algorithm
KW  -particle swarm
optimization
AB  - Forecasting is an important and widely popular topic in the research of system modeling. In this
study, we proposed a six 2-stage hybrid prediction models, wherein Laplacian Score (LS), Multi Cluster based
Feature Selection (MCFS), Correlation Based feature Selection (CBS) is used to construct Stage-1, followed by
invoking Adaptive Network based Fuzzy Inference System (ANFIS) trained by Genetic Algorithm (GA), Particle
Swarm Optimization (PSO)(Stage-2). We tested our model with Hang Seng Index (HSI) data and TAIEX stock
market transaction data from 1998-2006. The results compared with the existing models in the literature, the
comparison shows that the proposed model LS+ANFIS+GA outperformed the listing models in terms of both
of Root Mean Squared Error (RMSE) and Theil’s U statistic.
ER  -