Enhancing Forecast Accuracy: The Impact of Data Transformation in Time Series Models

Data transformations are an important tool for improving the accuracy of forecasts from time series models. Historically, the impact of transformations have been evaluated on the forecasting performance of different parametric and nonparametric forecasting models. However, researchers have overlooked the evaluation of this factor in relation to the nonparametric forecasting model of Singular Spectrum Analysis (SSA). In this paper, we focus entirely on the impact of data transformations in the form of standardisation and logarithmic transformations on the forecasting performance of SSA when applied to 100 different datasets with different characteristics. Our findings indicate that data transformations have a significant impact on SSA forecasts at particular sampling frequencies.

Additivity rule, isothermal and non-isothermal transformations on the basis of an analytical transformation model

Accurate retail sales forecasting is crucial for understanding customer demands, handling inventories, and optimizing business strategies. Conventional forecasting approaches struggle to take into consideration both linear and nonlinear transformations in the time series data, limiting the model's adaptability. Additionally, an effective business strategy is essential for improving overall company revenue, depending on insights from precise forecasting. In order to address these shortcomings, the Meta-Learning Enhanced Learnable Long Short-Term Memory network (Meta-LLSTM) is proposed for effective retail sales forecasting and generalization. The model applies meta learning for quickly adapting with enhanced forecasting under diverse operational conditions. On the contrary, the Multiple-Parameter Exponential Linear Unit (MPELU) introduces learnable parameters, enabling the model to handle both linear and nonlinear transformations to effectively forecast retail sales in business strategies. To improve retail sales forecasting, the following support modules are used in this study: Recency, Frequency, Monetary and Diversity (RFMD) to analyze customer sales information, K-means based customer segmentation, and Adaptive Inventory Correction (AIC) for information on inventories. Additionally, the metrics of Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Coefficient of Determination (R-squared), Mean Absolute Percentage Error (MAPE) and Symmetric Mean Absolute Percentage Error (SMAPE) are used to evaluate the Meta-LLSTM. The proposed model achieves an RMSE of 1.003 which is less than that of the state-of-the-art classifiers such as Auto Encoder (AE), Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU) and Recurrent Neural Network (RNN). Specifically, the RMSE of Meta-LLSTM is 16.97% less than the state of art approach (RNN), rendering it more effective than the existing models.

Data transformation, the process of preparing raw data for effective visualization, is one of the key challenges in information visualization. Although researchers have developed many data transformation techniques, there is little empirical study of the general impact of data transformation on visualization. Without such study, it is difficult to systematically decide when and which data transformation techniques are needed. We thus have designed and conducted a two-part empirical study that examines how the use of common data transformation techniques impacts visualization quality, which in turn affects user task performance. Our first experiment studies the impact of data transformation on user performance in single-step, typical visual analytic tasks. The second experiment assesses the impact of data transformation in multi-step analytic tasks. Our results quantify the benefits of data transformation in both experiments. More importantly, our analyses reveal that (1) the benefits of data transformation vary significantly by task and by visualization, and (2) the use of data transformation depends on a user's interaction context. Based on our findings, we present a set of design recommendations that help guide the development and use of data transformation techniques.

Artificial Neural Networks (ANNs) have been around for a while, and as technology has progressed, more people can have now access to Graphical Processing Units (GPUs), Tensor Processing Units (TPUs), and complex architectures. These days, deep neural networks are of the utmost significance in pattern recognition. One special application of the ANNs is the sequence classification and prediction. A special type of neural network with the capacity to remember patterns along with the temporal aspects have been widely used, they are the recurrent Neural Networks (RNNs). The Long Short Term Memory Networks (LSTMs) are improved versions of RNN with a better dealing of vanishing gradient problems. In this chapter, we discuss an LSTMs with their regular implementation as well as time distributed and bidirectional implementations for the purpose of sequence prediction.  Every day, new information about the COVID-19 pandemic's effects is released, and people all over the world are still dealing with its aftermath. Long Short-Term Memory (LSTM) networks are trained with this data in order to predict              estimates of the global impact of the COVID-19 pandemic. The LSTM architectures are discussed and compared with a vanilla RNN and the results are presented here. The results show the LSTMs outperform RNNs when the mean absolute error is compared for all the models.

Introduction. PIWI-interacting RNAs are small and non-coding RNAs involved in gene regulation and transposable element repression, emerging as critical biomarkers and therapeutic targets in oncology. Advances in artificial intelligence, such as recurrent neural networks, long short-term memory networks, and graph convolutional networks, offer significant improvements in PIWI-interacting RNA detection.Objectives. To evaluate the performance of artificial intelligence models, including recurrent neural networks, long short-term memory, and graph convolutional networks, in detecting PIWI-interacting RNAs and assessing their implications for cancer diagnostics and prognosis. Materials and methods. A systematic review of 24 studies was conducted across PubMed, ScienceDirect, Scopus, and Web of Science, focusing on artificial intelligence-based approaches for PIWI-interacting RNA detection. Inclusion criteria were original articles published in English or Spanish using artificial intelligence models in clinical or experimental settings. Performance metrics such as accuracy, sensitivity, and specificity were analyzed. Results. Long short-term memory models achieved the highest overall accuracy (92.3%), followed by graph convolutional networks (91.4%), support vector machines (88%), and recurrent neural networks (85.7%). Sensitivity and specificity were also highest in long short-term memory (94% and 91%, respectively). Graph convolutional networks showed superior performance in identifying PIWI-interacting RNA-disease associations with complex datasets. Support vector machine models were effective in smaller datasets but exhibited scalability limitations.Conclusion. Artificial intelligence models, especially long short-term memory and graph convolutional networks, significantly enhance PIWI-interacting RNA detection, supporting their application in cancer diagnostics and personalized medicine. Future studies should refine these models, address dataset biases, and explore their integration into clinical workflows.

A thermal deformation monitoring system was developed in this study by applying the thermocouple sensors and capacitive displacement sensors, along with a Long Short Term Memory (LSTM) Network Model classifier, for the alignment turning system (ATS). An ATS can simultaneously provide the functions of measuring the centration error and dimensions of the lens cell in-line, and machining the lens barrel housing with reference to the lens optical axis. The ATS can manufacture precise lens cells, applied for optical metrology, high numerical aperture objective lenses, and lithography projection lenses. While rising temperature, the thermal error would occur on hydrostatic spindle which build in ATS. Therefore, the predetermined machining point would offset, thereby resulting in the machining error. In order to acquire the oil temperature of rotor and the relative thermal displacement between hydrostatic spindle and turret, the thermocouple sensors and capacitive displacement sensors were assembling on ATS. According to the measurement of oil temperature and relative displacement, the thermal deformation monitoring system of ATS hydrostatic spindle was established. Cause of the high resolution of capacitive displacement sensors, the more precise measurement values could be obtain so that the monitoring system would have higher accuracy. LSTM is a variant of Recurrent Neural Network (RNN) and could remember longer information changes than traditional RNN. The thermal deformation monitoring system with LSTM could be applied to compensate the thermal error to improve the workpiece quality in real-time, and also could save time and money of warming up centering machines in the future. Results shows that the mean square error (MSE) and RScore of forecasting thermal error is less than 0.0002 and higher than 0.997, which is highly accurate forecasting.

Traditional neural networks cannot accurately analyze the sentiment tendency of the text according to the context of sentences. To address this issue, this paper proposes a recurrent neural network (RNN) model combined with long short-term memory (LTSM) networks. Our method solves the problem that traditional text analysis methods cannot efficiently analyze the emotional tendency of numerous online shopping comments. Firstly, a tokenizer is used to divide sentences into words for emotional analysis, a stop vocabulary table is used to filter out meaningless high-frequency vocabulary, and the Word2Vec is used to vectorize words. Then, the model combining LTSM and RNN is used to capture the emotional features contained in online texts. Finally, the model is optimized to the best performance by choosing different iterations. The experimental results show that the accuracy and F1 value of the proposed model have reached a high rate. The proposed method combining LTSM and RNN can effectively extract the emotional features of Internet text.

<p>The operation optimization of interconnected reservoirs is crucial for effective water resources management. Therefore, a decision support tool for is developed based on the forecasts of natural inflows. Standard forecast procedures are often based on historical streamflows and hydrological modelling of flows using quantitative meteorological forecasts. In recent years, forecasting using deep learning methods and especially recurrent neural networks have gained attention. Compared to other approaches such as regression-based and time series models, artificial neural networks have proven to be more effective and flexible. We propose a long short-term memory network (LSTM) for forecasting inflow into reservoirs with a large watershed. It is trained with observed hourly streamflow and meteorological data and applicable to different forecast horizons. The novelty here is the inclusion of temperature, windspeed and snow into the forecast.</p><p>The Drin river cascade (11 830 km²) in Northern Albania was selected as a pilot hydraulic system, whereby the upper part of the Drin river basin covers also parts of North Macedonia, Kosovo and Montenegro. The cascade consists of three large dams in series. The reservoirs are primarily used for energy generation and, secondarily, for flood retention. The studied LSTM forecast horizons (6, 8, 12 hours; >12 hours) indicate that the Recurrent Neural Network provides a proper forecast of the natural inflows into the reservoir cascade and thus represents a valuable tool for the optimization of the operation of the Drin Cascade under multi-criteria conditions.</p>

Recent studies on channel estimation in wireless communication systems have focused on deep learning methods. Our primary contribution is based on the use of DenseNet121 hybrid with Random Forest (RF), Gated Recurrent Units (GRU), Long Short-Term Memory Networks (LSTM), and Recurrent Neural Networks (RNN) to improve the channel estimation and lower the error rate. In order to mitigate inter-symbol interference and map the datasets, this paper introduces M-quadrature amplitude modulation (16-QAM) and orthogonal frequency division multiplexing (OFDM), which is based on quadrature phase shift keying (QPSK). Additionally, the existence or lack of cyclic prefixes forms the basis of our simulation. Additionally, the suggested models are investigated using pilot samples 2, 4, 8, and 64. Labeled OFDM signal samples, where the labels match the signal received after applying OFDM and passing through the medium, are used to train the proposed models. The DenseNet121 functions as a powerful feature extractor to extract intricate spatial information from received signal data. Sequential models like as RNN, LSTM, and GRU are used to model temporal dependencies in the retrieved features. RF is also utilized to exploit non-linear relationships and interactions between features to further increase prediction accuracy and reduce bit error rate (BER). By comparing the models using key metrics like accuracy, bit error rate (BER), and mean squared error (MSE), superior performance is attained based on the DenseNet121_RNN_GRU_RF model. Additionally, the DLMs are assessed against traditional methods like minimal mean square error (MMSE) and least squares (LS). Using the DenseNet121_RNN_GRU_RF model indicates a considerable gain over alternative architectures, with an improvement of 36.3% over DensNet121-RNN-LSTM-RF, according to a comparison of the suggested models without cyclic prefix for OFDM_QPSK. The improvement in percentages of roughly 63.3% over DensNet121-RNN-LSTM, 68.18% over DensNet121-GRU, 72.7% over DensNet121-LSTM, and 86.3% is the improvements of DenseNet121_RNN_GRU_RF over DensNet121-RNN are 86.3 and 72.7%, respectively, over DensNet121-GRU and DensNet121-LSTM. The DenseNet121_RNN_GRU_RF model performs better than the other models when compared to the suggested model with cyclic prefix for OFDM_QPSK. Compared to DenseNet121_RNN_LSTM_RF, the DenseNet121_RNN_GRU_RF model improves BER by about 45%. In contrast, the DenseNet121_RNN_GRU_RF model outperforms DenseNet121_RNN_LSTM by roughly 66.6%. It outperforms DenseNet121_GRU by 71.4%, DenseNet121_LSTM by 80.9%, and DenseNet121_RNN by 90.4%. Additionally, DenseNet121_RNN_GRU_RF shows a significant improvement over LS, requiring a 70% improvement over the LS approach. DenseNet121_RNN_GRU_RF outperforms the Minimum Mean Square Error (MMSE) by roughly 39.5%. Additionally, when using QPSK, higher pilot counts typically translate into lower MSE values. At MSE = 10-3, the improvement of employing 64 pilot bits over 8 pilot bits is approximately 12.1%. utilizing eight pilot bits improves performance by roughly 21.2% compared to utilizing two or four pilot bits. Performance is improved by approximately 18.9% at BER = 10-4 when there are eight pilots instead of four. Furthermore, there is a 13.8% improvement in accuracy from 8 to 64 pilots, indicating that more pilots can further increase accuracy. Finally, BER performance is greatly improved with additional pilots, as evidenced by the noteworthy 35.3% improvement between 4 and 64 pilots. For OFDM-QPSK, employing CP often results in an improvement of roughly 9% over not utilizing CP. Compared to the LS and MMSE models, the DenseNet121_RNN_GRU_RF model provides a significant BER improvement in terms of error rate reduction and computing time of 4.215 s. This suggests that the model's capacity to precisely estimate the channel and reduce bit errors has significantly improved.

This study aims to improve sentiment analysis accuracy and address overfitting challenges in deep learning models by developing a hybrid model based on Convolutional Neural Networks and Long Short-Term Memory Networks. The research methodology involved multiple stages, starting with preprocessing a dataset of 5,456 rows. This process included removing duplicate data, empty entries, and neutral sentiments, resulting in 2,685 usable rows. To overcome data quantity limitations, data augmentation expanded the training dataset from 2,148 to 10,740 samples. Data transformation was carried out using tokenization, padding, and embedding techniques, leveraging Word2Vec and GloVe to produce numerical representations of textual data. The hybrid model demonstrated strong performance, achieving a training accuracy of 99.51%, validation accuracy of 99.25%, and testing accuracy of 87.34%, with a loss value of 0.56. Evaluation metrics showed precision, recall, and F1-Score values of 86%, 87%, and 86%, respectively. The hybrid model outperformed individual models, including Convolutional Neural Networks (70% accuracy) and Long Short-Term Memory Networks (81% accuracy). It also surpassed other hybrid models, such as the multiscale Convolutional Neural Network-Long Short-Term Memory Network, which achieved a maximum accuracy of 89.25%. The implications of this study demonstrate that the hybrid model based on Convolutional Neural Networks and Long Short-Term Memory Networks effectively improves sentiment analysis accuracy while reducing the risk of overfitting, particularly in small or imbalanced datasets. Future research is recommended to enhance data quality, adopt more advanced embedding techniques, and optimize model configurations to achieve better performance.

This paper is dedicated to the development of recurrent neural networks in order to supplement the AI-based devices like microcontrollers and other mist computing systems. Due to the insignificant computational power of the edge deviсes the aim of the study is to design and analyze low complexity sequence models for a basic sensory time series forecasting on an example of univariate indoor temperature data. The description of data preparation and transformation followed by the models configuration via different architectures like simple LSTM and GRU is provided. To calculate an optimal set of hyper-parameters for the multiple neural network architectures a genetic algorithm has been implemented. The results of numerical experiments conducted for each model configuration consisting of both unidirectional and bidirectional cell connections are discussed. In addition to these studies the scheme of deploying the developed low-complexity models on STM32 microcontroller joined with the high-performance hub is proposed. Key words : edge computing, recurrent neural networks, time series, genetic algorithm.

Neural Network-based approaches have recently produced good performances in Natural language tasks, such as Supertagging. In the supertagging task, a Supertag (Lexical category) is assigned to each word in an input sequence. Combinatory Categorial Grammar Supertagging is a more challenging problem than various sequence-tagging problems, such as part-of-speech (POS) tagging and named entity recognition due to the large number of the lexical categories. Specifically, simple Recurrent Neural Network (RNN) has shown to significantly outperform the previous state-of-the-art feed-forward neural networks. On the other hand, it is well known that Recurrent Networks fail to learn long dependencies. In this paper, we introduce a new neural network architecture based on backward and Bidirectional Long Short-Term Memory (BLSTM) Networks that has the ability to memorize information for long dependencies and benefit from both past and future information. State-of-the-art methods focus on previous information, whereas BLSTM has access to information in both previous and future directions. Our main findings are that bidirectional networks outperform unidirectional ones, and Long Short-Term Memory (LSTM) networks are more precise and successful than both unidirectional and bidirectional standard RNNs. Experiment results reveal the effectiveness of our proposed method on both in-domain and out-of-domain datasets. Experiments show improvements about (1.2 per cent) over standard RNN.

The failure of water chiller will cause a series of problems such as increasing energy consumption, decreasing comfort of users and decreasing life of equipment. The fault diagnosis of the chiller has an important role in the stable operation of the air conditioner. In order to improve the fault diagnosis level of water chillers and solve the problems of difficulty in determining the parameters of neural network and great randomness, this paper presents a fault diagnosis method of long short-term memory (LSTM) network based on whale optimization algorithm (WOA) optimization. The fault diagnosis performance was verified by RP-1043 dataset, and compared with the results of recurrent neural network (RNN) and BP neural network. The results show that the WOA-LSTM method can effectively identify five typical chillers, and the diagnostic accuracy is as high as 99.27%. Compared with RNN and BP neural network, the proposed method is helpful to detect faults as early as possible and reduce the loss. It is especially obvious to improve the detection efficiency of small faults and complex faults.

A Recurrent Neural Network (RNN) is a specialized form of neural network that is adept at handling sequential data by retaining information from prior inputs. In contrast to conventional feedforward neural networks, RNNs incorporate loops in their architecture, allowing them to leverage data from previous time steps to affect the current output. This characteristic renders RNNs especially effective for applications that involve sequences, including time-series forecasting, natural language processing, and speech recognition. A fundamental component of RNNs is their hidden state, which acts as a dynamic memory that is refreshed with each incoming input. This allows RNNs to capture dependencies across time steps, which is crucial for understanding context in sequences. In language modeling, the interpretation of a word often relies on the words that come before it, a task that Recurrent Neural Networks (RNNs) handle well. However, RNNs struggle with issues like vanishing gradients, which hinder their ability to capture long-range dependencies. To overcome this, models such as Long Short-Term Memory (LSTM) networks and Gated Recurrent Units (GRUs) were introduced. These models incorporate gates that regulate the flow of information, allowing them to better learn long-term dependencies. RNNs remain a powerful tool for working with sequential data, facilitating the modeling of temporal relationships, but their effectiveness depends on careful design and optimization. Research significance: Recurrent Neural Networks (RNNs) hold significant research value because of their capacity to simulate temporal and sequential data, which is essential in many fields. They are frequently employed in natural language processing for tasks such as sentiment analysis, language translation, and text generation. In time-series analysis, RNNs enable accurate forecasting in finance, healthcare, and climate modeling. They also are essential in speech recognition and video processing, handling dependencies across time steps. Research focuses on improving RNNs, addressing challenges like vanishing gradients, and enhancing efficiency through architectures like LSTMs and GRUs, solidifying their relevance in advancing AI and machine learning applications. Methodology: A technique for analyzing the relationships between several variables, particularly in situations when data is limited or unclear, is called gray relational analysis, or GRA. In order to comprehend the relationships between variables, it evaluates how similar or different they are. GRA aids decision-makers in identifying critical factors, prioritizing actions, and improving processes in complex fields like engineering, finance, and management. By converting both qualitative and quantitative data into gray numbers, GRA addresses uncertainty and provides valuable insights for problem-solving, decision-making, and performance improvement, leading to more informed and effective strategies. Alternative taken as Simple RNN, LSTM, GRU, Bidirectional RNN, Deep RNN, Vanilla RNN, Echo State Network, Attention-based RNN, Transformer RNN, GRU with Attention. Evaluation preference taken as Prediction Accuracy, Model Robstness, Learning Efficiency, Training Time, Complexity. Attention-based RNN has the lowest score, Deep RNN has the highest rank, according to the results.