A Feature Fusion Based Forecasting Model for Financial Time Series

Abstract

Predicting the stock market has become an increasingly interesting research area for both researchers and investors, and many prediction models have been proposed. In these models, feature selection techniques are used to pre-process the raw data and remove noise. In this paper, a prediction model is constructed to forecast stock market behavior with the aid of independent component analysis, canonical correlation analysis, and a support vector machine. First, two types of features are extracted from the historical closing prices and 39 technical variables obtained by independent component analysis. Second, a canonical correlation analysis method is utilized to combine the two types of features and extract intrinsic features to improve the performance of the prediction model. Finally, a support vector machine is applied to forecast the next day's closing price. The proposed model is applied to the Shanghai stock market index and the Dow Jones index, and experimental results show that the proposed model performs better in the area of prediction than other two similar models.

Forecast Model Analysis for the Morbidity of Tuberculosis in Xinjiang, China

“…These measures have been used by many researchers to compare the accuracy of their models with other known models [30, 31, 32, 33, 34]. …”

Gradient Boost with Convolution Neural Network for Stock Forecast

“…Following [7,10,1,6], the results were evaluated by "MAPE", "Theil U" and "linear correlation between prediction and real prices" (use R to denote). These indicators are denoted as follows:…”

Soft computing and bioinspired computing techniques for stock market prediction-a comprehensive survey

“…The papers surveyed in this area focus on predicting the closing price of a single stock or multiple stock market values. However, some authors concentrate on predicting multiple stock values (Ajith et al [42]; Myungsook et al [24]; Ritanjali et al [70]; Tsung-Jung et al [26];Tarek et al [36];Savinderjit et al [35];Guo et al et al [65]). All surveyed paper is classified based on the number of inputs, input data size, pre-processing method, implemented technique, learning method, membership function, and training and test data sets.…”

“…Ina khanadelwal et al [69] study the Indian mining data and Shanghai stock market is examined by Xueshen et al [55] using twelve indicators. Tsung-Jung et al [26] uses fourteen indicators to model the Tai-wan stock and DIJA, Baba et al [46] study the Japanese stock using fifteen indicators, Taiwan stock is examined by Hsuan et al [39] using eighteen indicators, Guo et al et al [65]) uses twentynine indicators to forecast the Shanghai stock and DIJA. Some papers do not focus on particular indicators or other input variables, but uses price index, SA, news and fundamental variables.…”

“…DWT-ARIMA is developed by Jatinder et al [72].Swarnava et al [56] have presented a combined model DWT-BPNN .LM algorithm is used to train the developed model. In 2014, Guo et al [65] integrated ICA-CCR-SVR to increase the prediction accuracy. Table 4 presents the prediction model developed by researchers.…”