Simulation and Empirical Studies of Long Short-Term Memory Performance to Deal with Limited Data

Rainfall is the height of rainwater that falls on a flat area, assuming it doesn't evaporate, doesn't seep, and doesn't flow. Rain levels are measured in mm (millimeters). The target of the research being conducted is in Badung Regency, Bali because Bali is a tourist area that is often visited by tourists and from Indonesian itself, so predictions of meteorology, such as rainfall will greatly impact tourism. In this test, predictions use the Long Short Term Memory (LSTM) method, using daily weather data from the BMKG from 2010 to 2020 as training data and daily weather data for 2021 as prediction data. Based on the test results above, the results show that the two LSTM tests with LSTM Model 128.64 and LSTM Model 64.32 have low MAE and MAPE error values. From First Scenario, the Mean Absolute Error (MAE) value is 8.97246598930908 and Mean Absolute Percentage Error (MAPE) value is 1.7657206683278308%. From Second Scenario, the Mean Absolute Error is 9.706669940783014 and Mean Absolute Percentage Error is 1.9028466692362323%. From the MAE and MAPE values obtained in these two scenarios, it can be proven that from the evaluation results of Rainfall predictions in Badung Regency, Bali, the predictions can be said to be very accurate because they have an error value of less than 10.

Incorporating zero-carbon emission sources of energy into the electric grid is essential to meet the growing energy needs in public and industrial sectors. Smart grids, with their cutting-edge sensing and communication technologies, provide an effective approach to integrating renewable energy resources and managing power systems efficiently. Improving solar energy efficiency remains a challenge within smart grid infrastructures. Nonetheless, recent progress in artificial intelligence (AI) techniques presents promising opportunities to improve energy production control and management. In this study, initially, we employed two different Machine learning (ML) models: Recurrent Neural Network (RNN) and Long Short Term Memory (LSTM), to forecast solar power plant parameters. The analysis revealed that the LSTM model performed better than RNN in terms of Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Mean Squared Error (MSE). Following a review of the LSTM model’s graphical results, it was further enhanced by combining Autoencoder with LSTM, creating the Autoencoder LSTM (AELSTM) model. Next, a new hybrid model was introduced: Convolutional Neural Network-Autoencoder Long Short-Term Memory (HCAELSTM), designed to boost prediction accuracy. These models were trained on a one-year real-time solar power plant dataset for training and performance assessment. Ultimately, the hybrid HCAELSTM model surpassed the AELSTM model in terms of MAPE, MAE, and MSE. It excelled in MAPE scores for Daily Power Production, Peak Grid Power Production, and Solar Radiance, achieving low scores of 1.175, 2.116, and 1.592 respectively, demonstrating superior accuracy. The study underscores the importance of AI and ML, in particular, the hybrid model HCAELSTM, in enhancing the smart grid’s ability to integrate renewable energy sources. The hybrid model excels at accurately forecasting key measurements, improving solar power generation efficiency within the smart grid system which also plays a key role in the broader shift toward the fourth energy revolution.

The increasing use of cryptocurrencies has changed the dynamics of investment, presenting both opportunities and challenges for investors. Although various studies have compared the performance of Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) in predicting financial asset prices, there are still differences in results regarding which model is superior. Therefore, this study aims to compare the performance of LSTM and GRU in predicting Ethereum prices using a hyperparameter tuning approach. The data used is historical data of Ethereum (ETH) shares from 2020 to 2025. The research methodology includes data preprocessing using Min-Max scaling, model development with various layer configurations, and comprehensive evaluation using several performance metrics. The results show that the GRU Model provides superior performance with a lower Root Mean Squared Error (RMSE) of 0.0234 and Mean Absolute Error (MAE) of 0.0168, compared to LSTM's RMSE of 0.0265 and MAE of 0.0193. While LSTM exhibits a slightly better Mean Absolute Percentage Error (MAPE) of 18.08% compared to GRU at 18.17%, the GRU model achieves a higher R² Score of 0.9442 compared to LSTM at 0.9282. Visual analysis of the prediction patterns and residual distributions further demonstrates GRU’s more consistent and accurate performance in capturing Ethereum price movements. These findings suggest that while both models are effective for cryptocurrency price prediction, GRU offers slightly better overall performance and stability, especially in maintaining consistent prediction accuracy across different market conditions.

This paper proposes a model called X-LSTM-EO, which integrates explainable artificial intelligence (XAI), long short-term memory (LSTM), and equilibrium optimizer (EO) to reliably forecast solar power generation. The LSTM component forecasts power generation rates based on environmental conditions, while the EO component optimizes the LSTM model's hyper-parameters through training. The XAI-based Local Interpretable and Model-independent Explanation (LIME) is adapted to identify the critical factors that influence the accuracy of the power generation forecasts model in smart solar systems. The effectiveness of the proposed X-LSTM-EO model is evaluated through the use of five metrics; R-squared (R2), root mean square error (RMSE), coefficient of variation (COV), mean absolute error (MAE), and efficiency coefficient (EC). The proposed model gains values 0.99, 0.46, 0.35, 0.229, and 0.95, for R2, RMSE, COV, MAE, and EC respectively. The results of this paper improve the performance of the original model's conventional LSTM, where the improvement rate is; 148%, 21%, 27%, 20%, 134% for R2, RMSE, COV, MAE, and EC respectively. The performance of LSTM is compared with other machine learning algorithm such as Decision tree (DT), Linear regression (LR) and Gradient Boosting. It was shown that the LSTM model worked better than DT and LR when the results were compared. Additionally, the PSO optimizer was employed instead of the EO optimizer to validate the outcomes, which further demonstrated the efficacy of the EO optimizer. The experimental results and simulations demonstrate that the proposed model can accurately estimate PV power generation in response to abrupt changes in power generation patterns. Moreover, the proposed model might assist in optimizing the operations of photovoltaic power units. The proposed model is implemented utilizing TensorFlow and Keras within the Google Collab environment.

This study uses long short term memory (LSTM) modeling to predict time series data on the price of fertilizer raw materials, namely prilled urea, granular urea, ammonium sulphate((NH4)2SO4), ammonia (NH3), diammonium phosphate((NH4)2HPO4 ), phosphoric acid (H3PO4), phosphate rock (P2O5), NPK 16-16-16, potash, sulfur, and sulfuric acid (H2SO4). Predictions are made based on data that existed in the past using the long short term memory method, which is a derivative of the recurrent neural network. Carry out the evaluation process by looking at the root mean square error (RMSE) and mean absolute percentage error (MAPE) of the model that has been created. The results obtained are quite good, as seen from the root mean square error (RMSE) and mean absolute percentage error (MAPE) which are close to 0 and not too high. Sulfur raw material got the smallest root mean square error (RMSE) with a score of 0.053 and diammonium phosphate raw material got the smallest mean absolute percentage error (MAPE) evaluation value with 2.3%, while the largest value was for the root mean square error (RMSE) of raw materials. Phosphoric acid fertilizer raw material with a value of 22,979 and the largest mean absolute percentage error (MAPE) comes from sulfuric acid fertilizer raw material with a value of 9.180%.

Forecasting cryptocurrency prices using Recurrent Neural Network and Long Short-term Memory

BackgroundAccurate and reliable predictions of infectious disease can be valuable to public health organizations that plan interventions to decrease or prevent disease transmission. A great variety of models have been developed for this task. However, for different data series, the performance of these models varies. Hepatitis E, as an acute liver disease, has been a major public health problem. Which model is more appropriate for predicting the incidence of hepatitis E? In this paper, three different methods are used and the performance of the three methods is compared.MethodsAutoregressive integrated moving average(ARIMA), support vector machine(SVM) and long short-term memory(LSTM) recurrent neural network were adopted and compared. ARIMA was implemented by python with the help of statsmodels. SVM was accomplished by matlab with libSVM library. LSTM was designed by ourselves with Keras, a deep learning library. To tackle the problem of overfitting caused by limited training samples, we adopted dropout and regularization strategies in our LSTM model. Experimental data were obtained from the monthly incidence and cases number of hepatitis E from January 2005 to December 2017 in Shandong province, China. We selected data from July 2015 to December 2017 to validate the models, and the rest was taken as training set. Three metrics were applied to compare the performance of models, including root mean square error(RMSE), mean absolute percentage error(MAPE) and mean absolute error(MAE).ResultsBy analyzing data, we took ARIMA(1, 1, 1), ARIMA(3, 1, 2) as monthly incidence prediction model and cases number prediction model, respectively. Cross-validation and grid search were used to optimize parameters of SVM. Penalty coefficient C and kernel function parameter g were set 8, 0.125 for incidence prediction, and 22, 0.01 for cases number prediction. LSTM has 4 nodes. Dropout and L2 regularization parameters were set 0.15, 0.001, respectively. By the metrics of RMSE, we obtained 0.022, 0.0204, 0.01 for incidence prediction, using ARIMA, SVM and LSTM. And we obtained 22.25, 20.0368, 11.75 for cases number prediction, using three models. For MAPE metrics, the results were 23.5%, 21.7%, 15.08%, and 23.6%, 21.44%, 13.6%, for incidence prediction and cases number prediction, respectively. For MAE metrics, the results were 0.018, 0.0167, 0.011 and 18.003, 16.5815, 9.984, for incidence prediction and cases number prediction, respectively.ConclusionsComparing ARIMA, SVM and LSTM, we found that nonlinear models(SVM, LSTM) outperform linear models(ARIMA). LSTM obtained the best performance in all three metrics of RSME, MAPE, MAE. Hence, LSTM is the most suitable for predicting hepatitis E monthly incidence and cases number.

This study presents a novel, integrated deep-learning framework named 2LE-BO-DeepTrade for stock closing price prediction. This framework combines 2LE-ICEEMDAN denoising, deep learning models tuned with Bayesian optimization, and a piecewise linear representation (PLR)-based trading strategy. The framework utilizes the model that provides the highest accuracy among optimized long short-term memory (LSTM), long short term memory with batch normalization (LSTM-BN), and gated recurrent unit (GRU) models on data preprocessed with the 2LE-ICEEMDAN denoising method. The model’s performance is comprehensively evaluated using both statistical metrics and a PLR-based trading strategy specifically developed for this study. Experimental studies were conducted on AKBNK, MGROS, KCHOL, THYAO, and ULKER stocks, which are traded on Borsa Istanbul and represent different sectors. During the denoising phase, noise in the stock prices was successfully removed, and noiseless intrinsic mode functions (IMFs) were obtained. The optimal model and hyperparameters for each IMF component were determined using Bayesian optimization, significantly improving prediction accuracy. The model within this framework, characterized by its optimized yet simple structure, demonstrated superior predictive performance compared to the more complex ICE2DE-MDL model in the literature. When compared to ICE2DE-MDL, the 2LE-BO-DeepTrade model, across all tested stocks, reduced the average root mean square error (RMSE) value by 94.4%, the average mean absolute error (MAE) value by 93.6%, and the average mean absolute percentage error (MAPE) value by 37.4% while increasing the average R2 value by 1.1%. Furthermore, the PLR-based trading strategy, specifically developed for this study, generated “Buy” and “Sell” signals, exhibiting a remarkably superior financial performance to a passive investment strategy. Across all considered stocks, the PLR-based strategy yielded, on average, 66 times more profit than the passive approach. These findings substantiate that the proposed integrated deep learning-based stock forecasting framework can significantly enhance the accuracy of stock market predictions and the returns of trading strategies.

The retail company PT Terang Abadi Raya has a solid commitment to supporting distributors of LED lights and electrical equipment who have joined them, helping to spread their products widely in various regions. To face increasingly intense market competition, it is essential to produce high-quality products to win the competition and meet consumer demands. To achieve this, efficient production planning is necessary. The Convolutional Long Short-Term Memory (C-LSTM) method is used in this study to forecast product sales at PT Terang Abadi Raya. The research results show that C-LSTM has the potential to predict sales effectively. Evaluation is conducted using Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE). The calculations reveal that the smallest values are obtained at epoch 10, with an MAE of 0.1051 and a MAPE of 22% in the testing data. For the cable data, the smallest values are found at epoch 100, with an MAE of 0.0602 and a MAPE of 44% in the testing data. The Long Short-Term Memory (LSTM) method with ten neurons produces the most minor errors during training.

Beras merupakan makanan pokok masyarakat di Indonesia. Berdasarkan data Sistem Pemantauan Pasar dan Kebutuhan Pokok (SP2KP) Kementerian Perdagangan, harga beras di Indonesia terus meningkat sejak Agustus 2022 [1], hal ini berdampak pada tingkat kemiskinan, tingkat inflasi, dan stabilitas ekonomi. Oleh karena itu, pemerintah membutuhkan Solusi agar dapat menghindari dampak dari naik turunnya harga beras di Indonesia. Metode Long Short Term Memory (LSTM) memiliki kemampuan dalam menangani masalah ketergantungan jarak jauh pada data berurutan, sehingga cocok digunakan untuk memprediksi harga beras. Kombinasi parameter yang digunakan adalah metode normalisasi data, pembagian data, layer, batch size, epoch, dan learning rate. Hasil evaluasi model menggunakan MAPE (Mean Absolute Percentage Error) menunjukkan bahwa LSTM dapat memprediksi harga beras dengan akurasi 98.57% dan nilai MAPE 1.43%. Hasil prediksi terbaik didapatkan dengan menggunakan parameter metode normalisasi data StandardScaler, pembagian data validasi 10%, data latih 80%, data uji 10%, layer 2, batch size 4, epoch 40, dan learning rate 0.01, dengan pembagian data secara acak. Performa metode LSTM terbukti memiliki performa yang lebih baik. Untuk mengetahui performa LSTM, dilakukan perbandingan dengan metode ARIMA (Autoregressive Integrated Moving Average) dan RNN (Recurrent Neural Network). Nilai MAPE terbaik dari LSTM 1.43%, ARIMA 10.41%, dan RNN 3.53%. Adapun hasil akurasi terbaik LSTM ialah 98.57%, ARIMA 89.59%, dan RNN 96.47%. Kata Kunci— Long Short Term Memory, Autoregressive Integrated Moving Average, Recurrent Neural Network, Mean Absolute Percentage Error, prediksi, beras

Forecasting coal demand needs is important to minimize operational costs. Forecasting will help companies determine the right amount and time to order coal from suppliers. Research on coal forecasting in Indonesia generally uses a statistical approach and has not analyzed the performance of other forecasting models. This research aims to forecast coal demand using statistical and machine learning methods, namely ARIMA, Exponential Smoothing, Support Vector Regression (SVR), Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM). The evaluation methods used to analyze forecasting performance are Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE). The new coal demand data used is 1097 daily data taken from January 2021 to December 2022 in the form of a timeseries and is stationary which has been tested using Augmented Dickey-Fuller (ADF). The test results show that the ARIMA model has MAPE value of 5.11%, MAE 2.91 and R-Square 0.925, Exponential Smoothing MAPE 1.07%, MAE 0.55 and R-Square 0.997, SVR with MAPE value of 5.48%, MAE 3.16 and R-Square 0.88, RNN with MAPE value of 5.19%, MAE 2.91 and R-Square 0.896, LSTM with MAPE value of 4.83%, MAE 2.84 and R-Square 0.897. From the test results it was found that exponential smoothing had the smallest error values among the other models. With forecasting results that have a small error rate, it can help management in making decisions to minimize costs in coal ordering and warehouse management.

Sales estimates can be used to set product prices and increase expected profits. Flyover coffee shop Karanganyar does not have a methodical forecasting method to estimate and predict their need/demand for coffee beverage products. Two previous research that used Extreme Learning Machine (ELM) method in other predictions stated that ELM method has high accuracy and fast compilation time. Another research predicted jeans sales using the ARIMA model and produced an accuracy of 17.05% based on the MAPE (Mean Absolute Percentage Error) method. Menstrual cycle prediction using the Long Short-Term Memory (LSTM) method produces a MAPE value of 7.5%. Two advantages of ELM method from two previous research were used as the basis for selecting ELM method used in our study. To help predict sales of coffee beverage menus, this research utilized an artificial neural network method using ELM algorithm. ELM method consists of an input layer and an output layer connected through a hidden layer. Data used for the test was daily sales data for a month. Data used for this study consisted of 215 data samples. Daily sales data at the Flyover coffee shop were collected from June to December 2024. Based on the results and analysis of error values using MAPE method, an average error value was 8.274%. From comparison of original data results and prediction data, an average MAPE error value the best number of features and hidden neurons is 5.65%.

The clarification process is an important part of sugarcane production. This process is used for separating sediment and sugarcane juice by adding flocculant. The addition of quantity and concentration of flocculant directly affects the settling rate and turbidity of sugarcane juice. This paper proposes a model for forecasting quantity and concentration of flocculant by using Long Short-Term Memory (LSTM) Neural Network. Input data consists of green cane, burn cane, turbidity, and rainfall. Output data includes quantity and concentration of flocculant. Raw data was collected from top sugarcane factory and meteorological department in Thailand. The results are the forecast of the quantity and concentration of flocculant for one day in advance. The performance of LSTM is compared to the autoregressive integrated moving average (ARIMA), recurrent neural network (RNN), and gated recurrent unit (GRU) using root mean square error and mean absolute percent error. The result indicates that LSTM has the best performance. The forecast helps the operator in clarification process to prepare the flocculant.

An efficient management and better scheduling by the power companies are of great significance for accurate electrical load forecasting. There exists a high level of uncertainties in the load time series, which is challenging to make the accurate short-term load forecast (STLF), medium-term load forecast (MTLF), and long-term load forecast (LTLF). To extract the local trends and to capture the same patterns of short, and medium forecasting time series, we proposed long short-term memory (LSTM), Multilayer perceptron, and convolutional neural network (CNN) to learn the relationship in the time series. These models are proposed to improve the forecasting accuracy. The models were tested based on the real-world case by conducting detailed experiments to validate their stability and practicality. The performance was measured in terms of squared error, Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Mean Absolute Error (MAE). To predict the next 24 hours ahead load forecasting, the lowest prediction error was obtained using LSTM with R2 (0.5160), MLP with MAPE (4.97), MAE (104.33) and RMSE (133.92). To predict the next 72 hours ahead of load forecasting, the lowest prediction error was obtained using LSTM with R2 (0.7153), MPL with MAPE (7.04), MAE (125.92), RMSE (188.33). Likewise, to predict the next one week ahead load forecasting, the lowest error was obtained using CNN with R2 (0.7616), MLP with MAPE (6.162), MAE (103.156), RMSE (150.81). Moreover, to predict the next one-month load forecasting, the lowest prediction error was obtained using CNN with R2 (0.820), MLP with MAPE (5.18), LSTM with MAE (75.12) and RMSE (109.197). The results reveal that proposed methods achieved better and stable performance for predicting the short, and medium-term load forecasting. The findings of the STLF indicate that the proposed model can be better implemented for local system planning and dispatch, while it will be more efficient for MTLF in better scheduling and maintenance operations.

BackgroundInfluenza in mainland China results in a large number of outpatient and emergency visits related to influenza-like illness (ILI) annually. While deep learning models show promise for improving influenza forecasting, their technical complexity remains a barrier to practical implementation. Large language models, such as ChatGPT, offer the potential to reduce these barriers by supporting automated code generation, debugging, and model optimization.ObjectiveThis study aimed to evaluate the predictive performance of several deep learning models for ILI positive rates in mainland China and to explore the auxiliary role of ChatGPT-assisted development in facilitating model implementation.MethodsILI positivity rate data spanning from 2014 to 2024 were obtained from the Chinese National Influenza Center (CNIC) database. In total, 5 deep learning architectures—long short-term memory (LSTM), neural basis expansion analysis for time series (N-BEATS), transformer, temporal fusion transformer (TFT), and time-series dense encoder (TiDE)—were developed using a ChatGPT-assisted workflow covering code generation, error debugging, and performance optimization. Models were trained on data from 2014 to 2023 and tested on holdout data from 2024 (weeks 1‐39). Performance was evaluated using mean squared error (MSE), mean absolute error (MAE), and mean absolute percentage error (MAPE).ResultsILI trends exhibited clear seasonal patterns with winter peaks and summer troughs, alongside marked fluctuations during the COVID-19 pandemic period (2020‐2022). All 5 deep learning models were successfully constructed, debugged, and optimized with the assistance of ChatGPT. Among the 5 models, TiDE achieved the best predictive performance nationally (MAE=5.551, MSE=43.976, MAPE=72.413%) and in the southern region (MAE=7.554, MSE=89.708, MAPE=74.475%). In the northern region, where forecasting proved more challenging, TiDE still performed best (MAE=4.131, MSE=28.922), although high percentage errors remained (MAPE>400%). N-BEATS demonstrated the second-best performance nationally (MAE=9.423) and showed greater stability in the north (MAE=6.325). In contrast, transformer and TFT consistently underperformed, with national MAE values of 10.613 and 12.538, respectively. TFT exhibited the highest deviation (national MAPE=169.29%). Extreme regional disparities were observed, particularly in northern China, where LSTM and TFT generated MAPE values exceeding 1918%, despite LSTM’s moderate performance in the south (MAE=9.460).ConclusionsDeep learning models, particularly TiDE, demonstrate strong potential for accurate ILI forecasting across diverse regions of China. Furthermore, large language models like ChatGPT can substantially enhance modeling efficiency and accessibility by assisting nontechnical users in model development. These findings support the integration of AI-assisted workflows into epidemic prediction systems as a scalable approach for improving public health preparedness.