Forecasting Selected Colombian Shares Using a Hybrid ARIMA-SVR Model

Abstract

Forecasting future values of Colombian companies traded on the New York Stock Exchange is a daily challenge for investors, due to these stocks’ high volatility. There are several forecasting models for forecasting time series data, such as the autoregressive integrated moving average (ARIMA) model, which has been considered the most-used regression model in time series prediction for the last four decades, although the ARIMA model cannot estimate non-linear regression behavior caused by high volatility in the time series. In addition, the support vector regression (SVR) model is a pioneering machine learning approach for solving nonlinear regression estimation procedures. For this reason, this paper proposes using a hybrid model benefiting from ARIMA and support vector regression (SVR) models to forecast daily and cumulative returns of selected Colombian companies. For testing purposes, close prices of Bancolombia, Ecopetrol, Tecnoglass, and Grupo Aval were used; these are relevant Colombian organizations quoted on the New York Stock Exchange (NYSE).

Stock Price Prediction Using Arima Forecasting and LSTM Based Forecasting, Competitive Analysis

“…The linear regression-based ARIMA method does better using linear information, whereas the non-linear, elevated LSTM approach does effectively use share price time sequence information. [18] In latest days, period sequence research has shown that Long Short-term Memories (LSTM), an advancement of Recurring Neuronal Networks (RNN), performs effectively [17]. Because LSTM seems to have the property of growing as the cycle of events, the drawbacks of gradients vanishing and slope inflation inside the RNN models could be eliminated throughout the LSTM trained procedure of extended periods [15].…”

“…Overall precision of LSTM approach could surpass that of the ARIMA framework. The LSTM framework provides the highest precision when predicting the following day's share value, as per research by Lu et al who used a variety of algorithms to predict share values [18]. In contrast to LSTM, ARIMA involves a collection of characteristics (p, q, and d) that should be determined using information [13].…”

“…The learning procedure for a synthetic neuronal network entails recording past timeseries datasets, adjusting these using concealed network, and producing the outcome. Recurring Neuronal Networks (RNN) is a subdivision of ANN, and LSTM is a subdivision of RNN [18]. One sophisticated continuous neuronal network has considered Long Short-Term Memory (LSTM).…”

Demand Forecasting of Bento Considering the Product Popularity Estimation by Bayesian Rating System

“…In recent years, some studies have constructed forecasting models that combine two or more models. Rubio et al [7] proposed a hybrid model combining an ARIMA and support vector regression to forecast stock prices. Güven et al [8] developed a hybrid model that combined neural networks and support vector regression to forecast sales in the apparel industry.…”

“…Demand for bento is represented as a time series of sales-data. In research on demand forecasting for timeseries data, many methods have been proposed using statistical models and machine learning [2,3,4,5,6,7,8,9,10,11]. However, few studies have conducted demand forecasting to account for product popularity.…”