Nov 2026· Computer Science and Information Technology· 0 citations· 30 references
TL;DR
This study compares ARIMA, LSTM, and temporal fusion transformer (TFT) models across three applications and shows that TFT consistently achieved superior forecasting performance and demonstrated greater robustness to increasing missingness, while k-NN generally provided the most effective imputation performance across datasets.
Abstract
Time series forecasting plays a critical role in finance, healthcare, and energy applications, where accurate predictions support decision-making and operational efficiency. Traditional approaches such as autoregressive integrated moving average (ARIMA) perform well for linear patterns but often struggle with nonlinear and complex temporal dependencies found in real-world data. Although deep learning methods such as long short-term memory (LSTM) networks and Transformer-based models have shown promise, comprehensive evaluations across multiple domains and under missing-data conditions remain limited. This study compares ARIMA, LSTM, and temporal fusion transformer (TFT) models across three applications: stock price forecasting using SP 500 data, heart-rate prediction from electrocardiogram (ECG)-derived signals, and electricity demand forecasting using Pennsylvania–New Jersey–Maryland (PJM) power grid data. To evaluate robustness under realistic conditions, varying levels of missingness were introduced using missing completely at random (MCAR) and missing at random (MAR) mechanisms. Missing values were handled using forward fill, linear interpolation, and k-nearest neighbors (k-NN) imputation. Results show that TFT consistently achieved superior forecasting performance and demonstrated greater robustness to increasing missingness, while k-NN generally provided the most effective imputation performance across datasets.
In many fields, such as banking, healthcare, energy, and climate analysis, where precise predicting of future values is critical for making decisions, time series forecasting plays a critical role. With the quick development of machine learning methods, data-driven approaches have supplemented and frequently surpassed classic statistical models. This work provides a thorough analysis of deep learning architectures, contemporary machine learning algorithms, and traditional statistical models for time series forecasting. Critical analysis is done on important models including ARIMA, Support Vector Regression, Random Forests, Long Short-Term Memory (LSTM), and Transformer-based methods. The study also looks at evaluation criteria and benchmark datasets that are frequently used to compare performance. To illustrate how these models might be used in actual forecasting situations, a case study is provided. Problems including data quality, model interpretability, and computational complexity still exist despite tremendous advancements. Lastly, future directions are considered, such as explainable forecasting systems, automated machine learning, and hybrid models. An organized overview of contemporary developments and difficulties in time series forecasting is offered by this review.
Forecasting time series over long horizons is essential for proactive decision-making in many systems. Recent research has focused on transformer-based architectures, which capture long-range dependencies in sequential data. However, several studies show that simpler linear models can outperform transformers by avoiding overfitting during training. In this context, we present NeuroFlexMLP, a deep learning model for multivariate time series forecasting tasks. NeuroFlexMLP's key distinct feature is the adaptability to the diverse complexity of real-world time series, which is achieved, from the architecture standpoint, by adding non-linear residual blocks to a first linear block. This architectural design simplifies hyperparameter optimization, leading to accurate forecasts for various time series data types regardless of the lookback or prediction horizons, outperforming state-of-the-art (SOTA) models on challenging real-world datasets. Its Multi-Layer Perceptron (MLP) design ensures high computational efficiency, making it scalable for longer input sequences than transformer-based models. We validate NeuroFlexMLP for the LEO satellite beam hopping use case, where its lightweight design enables on-board deployment, and on state-of-the art AI datasets. Across all these benchmarks, NeuroFlexMLP achieves competitive accuracy over state-of-the-art models while providing an adaptive architecture that significantly reduces computational overhead. On the LEO beam hopping task, it achieves up to 35.9% MSE reduction over Informer, which translates into up to 28% lower provisioning cost under asymmetric cost models that penalize under-allocation more heavily than over-allocation.
P. F. Pérez, Claudio Fiandrino, Marco Fiore et al.· La Main· 0 citations
This study develops a rigorous, leakage-free forecasting framework for monthly Robusta coffee prices using historical observations from January 1975 to December 2025. A comprehensive set of explanatory variables is constructed from lagged coffee prices, moving averages, logarithmic returns, rolling volatility, and exogenous variables such as the Oceanic Niño Index (ONI), the U.S. Dollar Index, and Brent crude oil prices. To ensure methodological fairness, all predictors are generated exclusively from information available at the forecast origin, and all competing models are evaluated under a unified expanding-window walk-forward validation framework. Seven forecasting models are compared: Naïve, Exponential Smoothing (ETS), ARIMA, ARIMAX, Extreme Gradient Boosting (XGBoost), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU). Forecasting performance is evaluated using R2, RMSE, MAE, and MAPE, while Taylor diagrams and the Diebold–Mariano test are employed to assess model agreement and differences in predictive accuracy. The results show that XGBoost achieves the highest forecasting accuracy (R2 = 0.956, RMSE = 0.264), followed closely by the Naïve (R2 = 0.954, RMSE = 0.271) and ARIMA (R2 = 0.954, RMSE = 0.270) benchmarks, whereas ARIMAX and ETS provide comparable performance and the deep learning models (LSTM and GRU) produce substantially larger prediction errors. Feature importance analysis further indicates that the first lag of coffee price is the dominant predictor, accounting for approximately 94% of the predictive gain in XGBoost. Overall, the findings demonstrate that rigorous leakage-free validation is essential for reliable forecasting research and that, for monthly Robusta coffee prices, increased model complexity does not necessarily yield superior predictive performance.
Dler H Kadir, D. Khalil, Azhin M. Khudhur· Forecasting· 0 citations
Increasing air pollution in urban areas, particularly in DKI Jakarta, has made a reliable air quality prediction system increasingly essential for environmental control and public health risk management. The Air Quality Index (AQI) exhibits complex and fluctuating patterns, requiring forecasting methods capable of capturing both linear and non-linear. This study aims to conduct a comparative analysis of statistical, deep learning, and hybrid models for AQI forecasting using daily AQI data from Jakarta during the 2023–2025 period. The dataset includes polutant parameters such as , , , , , and CO. The proposed models consist of Autoregressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM), Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM), and a hybrid ARIMA-LSTM model. The research methodology includes data preprocessing, normalization using Min-Max Scaling, sequence generation using the sliding window approach, model training, and evaluation using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). The experimental results show that the CNN-LSTM model achieved the best forecasting performance with MAE, RMSE, and MAPE values of 4.54, 6.03, and 13.83%, respectively, followed closely by the LSTM model. Meanwhile, the ARIMA model produced the lowest performance, and the hybrid ARIMA-LSTM model did not outperform the standalone deep learning models. These findings indicate that deep learning approaches, particularly CNN-LSTM, are more effective in capturing the complex dynamics of urban air pollution data and have strong potential to support air quality forecasting and pollution control systems in Jakarta.
Mushliha, Nisrinah· IDEALIS : InDonEsiA journaL...· 0 citations
The Decomposed Recurrent Neural Network (DeRNN) is proposed, which decouples global trend modeling from local fluctuation extraction via an asymmetric dual-track architecture and exhibits superior robustness against noise and distribution shifts.
Shanyun Qian· Poster Volume 0008 The 2026...· 0 citations
Accurate prediction of financial time series is still a difficult problem as financial markets display high volatility, non-linearity and stochasticity. Traditional forecasting methods necessitate extensive domain knowledge in designing technical indicators for subsequent analysis, often resulting in the loss of intricate time dependencies. The goal of the present study is to propose a framework allowing for learning representations automatically from raw financial data that are informative in downstream forecasting tasks. The proposed framework, contrasting predictive coding (CPC), is based on self-supervised representation learning. The learned embeddings are applied to Linear Regression, Random Forest and LSTM to predict the next-day log returns of three major foreign exchange currency pairs: EUR/USD, GBP/USD and USD/JPY. Evaluating the Performance of CPC-Generated Representations and Conventional Handcrafted Features on Forecasting Models trained on Historical Market Data. The LSTM with CPC context embeddings produces the best overall performance with a drop in mean squared error of 18%, directional prediction accuracy of roughly 59%, and better risk-adjusted trading performance with Sharpe ratios above 0.7. Additionally, the outcomes of transfer learning experiments reveal that a CPC encoder trained using one currency pair efficiently generalizes to other currency pairs. The results indicate that self-supervised representation learning can serve as an effective and scalable substitute for manual feature engineering in finance time-series forecasting.
Muskan Pawar· International Journal of Ada...· 0 citations