Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Journal of Mathematics and Applied Statistics
Kalange, Dhananjay
Unknown Affiliation
Mathematics

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Time Series Forecasting in Finance Using Convolutional Neural Network (CNN) model Kalange, Dhananjay
Journal of Mathematics and Applied Statistics Vol. 3 No. 1 (2025): June 2025
Publisher : Yayasan Insan Literasi Cendekia (INLIC) Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35914/mathstat.v3i1.285

Abstract

This study investigates the effectiveness of a one-dimensional Convolutional Neural Network (1D-CNN) model in forecasting stock prices of selected companies listed on the Bombay Stock Exchange (BSE). Due to the highly volatile and non-linear nature of financial time series, traditional models like ARIMA often fail to capture hidden patterns. To address this challenge, this research employs a CNN model trained on historical stock data (2015–2023) from five prominent Indian companies: SBI, Reliance, TCS, Infosys, and HDFC Bank. The data pre-processing included handling missing values, applying Min-Max normalization, and using a sliding window of 60 days to predict the next day's closing price. The CNN model, structured with convolutional and pooling layers followed by a dense network, was trained and validated using 80/20 data splits. Model performance was evaluated using MAE, MSE, RMSE, and MAPE. Results revealed that the CNN model achieved MAPE values between 1.5% and 4.2%, demonstrating high accuracy. Compared to ARIMA and LSTM models, CNN provided competitive predictive performance with faster training time. Visual comparisons and performance metrics confirmed the model's ability to capture both upward and downward market trends effectively. This research contributes to the growing field of deep learning in financial forecasting and supports the utility of CNNs in modeling complex time-series data.
Co-Authors