Time Series Of Variables Research Articles

Short term forecasting of persistent air quality deterioration events in the metropolis of Sao Paulo

Air pollution is one of the main environmental problems in metropolitan areas. Negative impacts to human health are intensified when poor air quality conditions persist for many consecutive days, with gradual accumulation of pollutants in the atmosphere. Persistent air quality deterioration events are typically associated with occurrence of stagnant atmospheric conditions and reach the regional scale. In this study, data-driven models were developed to forecast the occurrence of persistent air pollution events of inhalable particulate matter (PM10) and ozone (O3) in the metropolis of Sao Paulo, Brazil. On average, 8 events per year were observed between 2005 and 2022, comprising 73 event days per year. The logistic regression method was used in a supervised learning framework. Daily timeseries of surface weather variables were used as predictors. In the case of PM10, a consistent long-term decrease in the number events impacted the model performance. The PM10 model benefited from the restriction of the training set to recent years, with a significant increase in the model accuracy despite the reduction in the volume of data. The final models correctly reproduced the seasonal distribution of events, with overall accuracies of 0.92 and 0.87 for O3 and PM10, respectively, in 2022. Despite the fact that persistent exceedance events are relatively rare, the models were able to detect 81% and 97% of the event days in 2022, respectively for O3 and PM10. Daily maximum temperature was an important predictor, increasing the event odds by 483% (O3) and 84% (PM10). The classification models developed in this study can successfully forecast the occurrence of regional air pollution events concerning both primary and secondary air pollutants, which have different drivers for accumulation in the atmosphere. The models require simple input data and low computational resources, aiming to stimulate future usage by the general public and decision-makers, in order to mitigate exposure to harmful air pollutant concentrations.

Semi-real-time decision tree ensemble algorithms for very short-term solar irradiance forecasting

Industrial activities are transitioning towards decarbonization, focusing on renewable energy sources, particularly photovoltaic solar energy. However, the inherent high variability of photovoltaic energy poses challenges. Some of them can be partially addressed by predicting electricity production, which in the case of photovoltaic solar energy is heavily based on solar irradiance prediction. Although extensive research has been conducted in this field, there is a noticeable gap in research regarding very short-term (intra-minute) forecasting under high-variability scenarios. In this proposal, real data from a photovoltaic solar plant in Alderville (Canada) were used to predict irradiance with a horizon of 15 and 30 s. The objective is to make this prediction in near-real time. To achieve this, we propose the use of machine learning algorithms based on decision tree ensembles, due to their low computational training cost and known effectiveness. On the other hand, we propose pre-processing the data through a temporal and spatial correlation analysis between measurements from different sensors. Feature selection analysis allows us to determine the direction of the wind and consequently identify the most relevant panels for model training. This preprocessing enhances the model retraining without the need for external information such as sky images or wind speed and direction on days with highly variable cloud cover. The presented methodology offers promising results with significantly reduced training times, demonstrating the suitability of this semi-online training approach for highly variable time series forecasting.

Data-driven approach for the detection of faults in district heating networks

At present, District Heating Networks (DHNs) are required to operate more and more reliably and efficiently in order to further save primary energy and reduce environmental impact. Thus, monitoring and diagnostic approaches are necessary to identify the typical faults that affect this type of systems (e.g., anomalous heat and pressure losses), with the final goal of optimizing DHN operation and management. To this aim, this paper presents a data-driven diagnostic methodology that exploits NARX (nonlinear autoregressive network with exogenous inputs) neural networks to simulate DHN healthy operation and a threshold-based criterion for fault detection and identification. The novel diagnostic methodology is tested for evaluating the health state of the DHN of the campus of the University of Parma (Italy), based on the availability of time series of measurable variables (mass flow rate, pressure, and temperature) for both the power plant and the end-users. Both single and multiple faults of anomalous heat losses and anomalous pressure losses, with different magnitudes and location, were artificially implanted in DHN pipes. The main novel contribution of this paper with respect to state-of-the-art literature relies on the development of a real-time simulation approach aimed to predict DHN future operation and detect abnormal deviations from normal operation. The methodology proves to correctly detect and identify all simulated faults related to heat and pressure losses, by also correctly estimating their magnitude even in the most challenging scenarios.

A Decadal Survey Without Analogs: Earth Observation Needs for a Warming World

AbstractSince 2007, the National Academy for Sciences Engineering and Medicine (NASEM) has recommended Earth Science research and investment priorities every 10 years. The Decadal Survey balances the continuation of essential climate variable time series against unmet measurement needs and new Earth Observations made possible by technological breakthroughs. The next survey (2027–2028, DS28) will be framed by a rapidly changing world, and it will be critical to anticipate the observational needs of the 2030s–2040s, a world increasingly dominated by climate extremes and a rapidly changing Earth system. Here, we highlight some of the changes that factor into a framework for the DS28.

Impact of atmospheric O3 and NO2 on the secondary sulfates in real atmosphere

As an important component of secondary aerosols, sulfate plays a crucial role in regulating atmospheric radiative balance and influencing the secondary formation of ozone (O3). In real atmosphere, atmospheric oxidants NO2 and O3 can promote the oxidation of SO2 to form sulfate (SO42−) through multiphase chemistry that occur at different time scales. Due to the combined impact of meteorology, pollution sources, atmospheric chemistry, etc., time-scale dependence of SO2-SO42− conversion makes the impact of NO2/O3 on it more complex. In this study, based on long-term time series (2013-2020) of air pollution variables from seven stations in Hong Kong, the Multifractal Detrended Cross-Correlation Analysis (MFDCCA) method has been employed to quantify the cross-correlations between SO2 and SO42− in real atmosphere at different time scales, for examining the time-scale dependence of SO2-SO42− conversion efficiency. Furthermore, the Pearson correlation analysis has been used to study the influence of NO2/O3 on SO2-SO42− conversion, and the regional and seasonal differences have been analyzed by considering factors such as meteorology, pollution sources, and regional transport. Changes in the main components of secondary aerosols are closely linked with the co-control of regional PM2.5 and O3. Therefore, the exploration of the impact of co-existing NO2/O3 gases on the secondary formation of sulfates in real atmosphere is significant.

Boosting crop classification by hierarchically fusing satellite, rotational, and contextual data

Accurate early-season crop type classification is crucial for the crop production estimation and monitoring of agricultural parcels. However, the complexity of the plant growth patterns and their spatio-temporal variability present significant challenges. While current deep learning-based methods show promise in crop type classification from single- and multi-modal time series, most existing methods rely on a single modality, such as satellite optical remote sensing data or crop rotation patterns. We propose a novel approach to fuse multimodal information into a model for improved accuracy and robustness across multiple crop seasons and countries. The approach relies on three modalities used: remote sensing time series from Sentinel-2 and Landsat 8 observations, parcel crop rotation and local crop distribution. To evaluate our approach, we release a new annotated dataset of 7.4 million agricultural parcels in France (FR) and the Netherlands (NL). We associate each parcel with time-series of surface reflectance (Red and NIR) and biophysical variables (LAI, FAPAR). Additionally, we propose a new approach to automatically aggregate crop types into a hierarchical class structure for meaningful model evaluation and a novel data-augmentation technique for early-season classification. Performance of the multimodal approach was assessed at different aggregation levels in the semantic domain, yielding to various ranges of the number of classes spanning from 151 to 8 crop types or groups. It resulted in accuracy ranging from 91% to 95% for the NL dataset and from 85% to 89% for the FR dataset. Pre-training on a dataset improves transferability between countries, allowing for cross- domain and label prediction, and robustness of the performances in a few-shot setting from FR to NL, i.e., when the domain changes as per with significantly new labels. Our proposed approach outperforms comparable methods by enabling deep learning methods to use the often overlooked spatio-temporal context of parcels, resulting in increased precision and generalization capacity.

Selection of best wavelet functions and decomposition levels for coupling with soft computing methods in estimating ETo in coastal and island regions

ABSTRACT Mass transfer-based models are among the extensively applied reference evapotranspiration (ET o ) estimation approaches that have been proven to provide reliable estimates under certain climatic and geographical conditions. A major issue with such approaches is the high variations in wind speed time series, which is a governing meteorological parameter in mass transfer-based models. few studies have focused on applying wavelet transform as a robust pre-processing technique for ETo estimation. A major question with coupling wavelet transforms with soft computing methods would be the proper choice of the wavelet function, vanishing moments and the decomposition levels. The present study aims at assessing different wavelet functions coupled with boosted regression tree (BT) models, i.e. wavelet-BT (WBT), in estimating ET o values (through mass transfer parameters) in both coastal and island regions, where it is supposed that wind speed variations are affected by sea–inland interactions and so has the sharp variations (unlike ET o ). The obtained results showed that the coupled WBT model could improve the single BT model results to a great extent, and meanwhile, the Daubechies wavelet function with six vanishing moments provided the most accurate results in all the locations. Further, the best simulation outcomes with each wavelet function were obtained when the maximum possible decomposition level was used with each function.

Using Machine Learning Techniques to Forecast Mehram Company's Sales: A Case Study

Sales forecasting, situated at the intersection of art and science, is critical for inspiring managers toward achieving profitable outcomes. Its precision sustains production levels and capital and plays a pivotal role in the company's and its leaders' overall success and career progression. In the context of Mahram Food Industries, the challenge arises from diverse investor perspectives and the impactful nature of numerous variables. To address this, a new sales forecasting algorithm has been introduced to enhance accuracy. The aim is to predict future sales through a comprehensive approach, leveraging technical analysis, time series modeling, machine learning, neural networks, and random forest techniques. The research methodology integrates various advanced techniques to improve sales forecasting for Mahram Food Industries. Technical analysis, time series modeling, machine learning, neural networks, and random forest methods are combined to create a robust framework. The focus is on predicting sales for a future period within the artificial intelligence-based machine learning domain. The study employs metrics such as Mean Absolute Deviation (MAD), MAD Percentage (MADP), and Mean Squared Error (MSE) to evaluate and compare the performance of the proposed neural network against traditional methods like multiple variable regression and time series modeling. The study's results highlight the superior performance of the neural network in sales forecasting for Mahram Food Industries. The Mean Absolute Deviation (MAD) for the neural network is 28.33, outperforming multiple variable regression (28.54) and time series modeling (29.45). Additionally, the neural network demonstrates a better MAD Percentage (MADP) with a value of 10.2%, surpassing the values associated with multiple variable regression (10.35%) and time series modeling (10.30%). The Mean Squared Error (MSE) further confirms the neural network's superiority with a value of 6452 compared to 6472 and 7865 for multiple variable regression and time series modeling, respectively. In conclusion, the study showcases the effectiveness of integrating advanced techniques, particularly the neural network, in enhancing the accuracy of sales forecasting for Mahram Food Industries. The comprehensive approach, combining technical analysis, time series modeling, machine learning, neural networks, and random forest, is a valuable strategy for predicting future sales. The superior performance of the neural network, as evidenced by lower MAD, MADP, and MSE values, suggests its potential for guiding informed decision-making in goal setting, hiring, budgeting, and other critical aspects of business management.

Accounting for trade patterns

We develop a quantitative framework for decomposing trade patterns. We derive price indexes that determine comparative advantage and the aggregate cost of living. If firms and products are imperfect substitutes, we show that these price indexes depend on variety, average appeal (including quality), and the dispersion of appeal-adjusted prices. We show that they are only weakly related to standard empirical measures of average prices. We find that 40 percent of the cross-section variation in comparative advantage, and 90 percent of the time-series variation, is accounted for by variety and average appeal, with less than 10 percent attributed to average prices.

A data decomposition and attention mechanism-based hybrid approach for electricity load forecasting

An accurate and reliable prediction of future energy patterns is of utmost significance for the smooth operation of several related activities such as capacity or generation unit planning, transmission network optimization, better resources availability, and many more. With the availability of historical load datasets through smart grid systems, artificial intelligence and machine learning-based techniques have been extensively developed for achieving the desired objectives. However, effectively capturing strong randomness and non-linear fluctuations in the load time-series remains a critical issue that demands concrete solutions. Considering this, the current research proposes a hybrid approach amalgamating data smoothing and decomposition strategy with deep neural models for improving forecasting results. Moreover, an attention mechanism is integrated to capture relevant portions of the time series, thus achieving the desired ability to capture long-term dependencies among load demand observations. This integration enhances the prediction and generalization capabilities of the proposed model. To validate the performance benefits achieved by the proposed approach, a comparative evaluation is conducted with state-of-the-art neural-based load series prediction models. The performance assessment is carried out on a novel real-world dataset of five southern states of India, and the superiority of the proposed in capturing load time-series variations is well observed and demonstrated in terms of several performance indicators.

Analysts’ extrapolative expectations in the cross-section

This paper examines extrapolative patterns of analysts’ expectations in the cross-section of firms. Using analysts’ target prices, I estimate the degree of extrapolative weighting capturing the relative weight analysts place on recent versus distant realized returns when forming their price expectation. I show considerable levels of extrapolation in the overall sample and on firm level. Results suggest considerable cross-sectional variation of extrapolation with valuation difficulty having a positive impact on the degree of extrapolative weighting. Furthermore, I construct a time-series of the degree of extrapolative weighting and argue that its time-series variation is also explained by valuation difficulty.

Evaluation of soft-computing techniques for pan evaporation estimation

Estimation of pan evaporation (Epan) can be useful in judicious irrigation scheduling for enhancing agricultural water productivity. The aim of present study was to assess the efficacy of state-of-the-art LSTM and ANN for daily Epan estimation using meteorological data. Besides this, the effect of static time-series (Julian date) as additional input variable was investigated on performance of soft-computing techniques. For this purpose,the models were trained, tested and validated with eight meteorological variables of 37 years by using preceding 1-, 3- and 5- days’ information. Data were partitioned into three groups as training (60%), testing (20%), and validation (20%) components. It was observed that the models performed well (best) with preceding 5-days meteorological information followed by 3-days and 1-day. However, all LSTMs simulated peak value of Epan was more accurate as compared to lower values. Meteorological data with julian date improved the performance of LSTMs (0.75<NSE 1; PBias< 10; KGE 0.75). The ANN trained using only meteorological data (preceding 5-days information) had better performance error statistics among all other ANNs and LSTMs with minimum MAE (0.68 to 0.86), RMSE (0.93 to 1.22), PBias (-0.73 to 2.44) and maximum NSE (0.83 to 0.84) and KGE (0.89 to 0.92). Overall, it was inferred that the forecasting of meteorological parameters using a few days preceding information along with Julian date as the time series variables resulted in better estimation of Epan for the study region.

Monitoring Flow-Forming Processes Using Design of Experiments and a Machine Learning Approach Based on Randomized-Supervised Time Series Forest and Recursive Feature Elimination.

The machines of WF Maschinenbau process metal blanks into various workpieces using so-called flow-forming processes. The quality of these workpieces depends largely on the quality of the blanks and the condition of the machine. This creates an urgent need for automated monitoring of the forming processes and the condition of the machine. Since the complexity of the flow-forming processes makes physical modeling impossible, the present work deals with data-driven modeling using machine learning algorithms. The main contributions of this work lie in showcasing the feasibility of utilizing machine learning and sensor data to monitor flow-forming processes, along with developing a practical approach for this purpose. The approach includes an experimental design capable of providing the necessary data, as well as a procedure for preprocessing the data and extracting features that capture the information needed by the machine learning models to detect defects in the blank and the machine. To make efficient use of the small number of experiments available, the experimental design is generated using Design of Experiments methods. They consist of two parts. In the first part, a pre-selection of influencing variables relevant to the forming process is performed. In the second part of the design, the selected variables are investigated in more detail. The preprocessing procedure consists of feature engineering, feature extraction and feature selection. In the feature engineering step, the data set is augmented with time series variables that are meaningful in the domain. For feature extraction, an algorithm was developed based on the mechanisms of the r-STSF, a state-of-the-art algorithm for time series classification, extending them for multivariate time series and metric target variables. This feature extraction algorithm itself can be seen as an additional contribution of this work, because it is not tied to the application domain of monitoring flow-forming processes, but can be used as a feature extraction algorithm for multivariate time series classification in general. For feature selection, a Recursive Feature Elimination is employed. With the resulting features, random forests are trained to detect several quality features of the blank and defects of the machine. The trained models achieve good prediction accuracy for most of the target variables. This shows that the application of machine learning is a promising approach for the monitoring of flow-forming processes, which requires further investigation for confirmation.

The myth of tightening credit rating standards in the market for corporate debt

This study revisits the existing evidence of a downward trend in credit rating standards indicating that Credit Rating Agencies (CRAs) have become more conservative over time. We find that the time-series variation in the proxy for rating standards is mostly driven by the market-based variables in the model, specifically market capitalization and idiosyncratic volatility. We examine an alternative specification of the model, incorporating risk characteristics of rated firms relative to those of the average firm in the economy, and find it to have a higher explanatory power. Most importantly, we find little evidence of increased conservatism over time, in contrast to prior studies.

A variance-upscaling quantile mapping method for gridded precipitation bias correction

Accurate estimation of precipitation in mountainous regions remains a challenge. Quantile Mapping (QM) is a method used extensively to correct gridded precipitation from numerical models or satellite remote sensing. The method can match the intensity histogram of a given gridded precipitation to that of the observation. However, the widely existing oversmoothed effect, a precipitation time series variability representativeness error, in observational gridded datasets may be inherited in the corrected results, impacting the assessment of extreme precipitation events. To address this issue, this study introduces a variance-upscaling QM method (VUQM). Unlike previous methods that obtain frequency parameters by interpolating rain gauge data, the new method creates a model to transform the precipitation time series variance at the rain gauge point level to the model's grid level. When compared to correct ERA5-Land reanalysis precipitation, the results show that the variance using the VUQM is 1.6 times larger than that obtained using QM alone, and it is 2 times larger in the 95th percentile. This indicates that the new method significantly reduces the oversmoothed effect and preserves precipitation variation. Random simulations show that low station density only introduces uncertainty, not theoretical bias in VUQM. In contrast, the variance biases in gridded datasets constructed from areal average or interpolation cannot be avoided, particularly in regions with significant spatial heterogeneity. This suggests that VUQM is well-suited for use in the Qinghai-Tibetan Plateau (QTP) compared with the old methods. Overall, this method sheds some light on mountainous precipitation bias correction.

An end-to-end machine learning approach with explanation for time series with varying lengths

AbstractAn accurate prediction of complex product quality parameters from process time series by an end-to-end learning approach remains a significant challenge in machine learning. A special difficulty is the application of industrial batch process data because many batch processes generate variable length time series. In the industrial application of such methods, explainability is often desired. In this study, a 1D convolutional neural network (CNN) algorithm with a masking layer is proposed to solve the problem for time series of variable length. In addition, a novel combination of 1D CNN and class activation mapping (CAM) technique is part of this study to better understand the model results and highlight some regions of interest in the time series. As a comparative state-of-the-art unsupervised machine learning method, the One-Nearest Neighbours (1NN) algorithm combined with dynamic time warping (DTW) was used. Both methods are investigated as end-to-end learning methods with balanced and unbalanced class distributions and with scaled and unscaled input data, respectively. The FastDTW and DTAIDistance algorithms were investigated for the DTW calculation. The data set is made up of sensor signals that was collected during the production of plastic parts. The objective was to predict a quality parameter of plastic parts during production. For this research, the quality parameter will be a difficult or only destructively measurable parameter and both methods will be investigated for their applicability to this prediction task. The application of the proposed approach to an industrial facility for producing plastic products shows a prediction accuracy of 83.7%. It can improve the reverence method by approximately 1.4%. In addition to the slight increase in accuracy, the CNN training time was significantly reduced compared to the DTW calculation.

Exploring socioeconomic similarity-inequality: a regional perspective

AbstractSocioeconomic variables have been studied in many different contexts. Considering several socioeconomic variables as well as using the standard series clustering technique and the Ward’s algorithm, we rank the countries in the world and evaluate the similarity and inequality between geographic areas. Various relationships between variables are also identified. Additionally, since the Gini coefficient is one of the most frequently used metrics to measure economic inequality, with a global scope, we model this coefficient utilising machine learning techniques. 16 exploratory variables are utilised, which pertain to the health (9), economic (2), social labour protection (4) and gender (1) fields. International repositories that include time series of variables referred to these domains as well as education and labour market fields are used.

On the use of fractional calculus to improve the pulse arrival time (PAT) detection when using photoplethysmography (PPG) and electrocardiography (ECG) signals.

The pulse arrival time (PAT) has been considered a surrogate measure for pulse wave velocity (PWV), although some studies have noted that this parameter is not accurate enough. Moreover, the inter-beat interval (IBI) time series obtained from successive pulse wave arrivals can be employed as a surrogate measure of the RR time series avoiding the use of electrocardiogram (ECG) signals. Pulse arrival detection is a procedure needed for both PAT and IBI measurements and depends on the proper fiducial points chosen. In this paper, a new set of fiducial points that can be tailored using several optimization criteria is proposed to improve the detection of successive pulse arrivals. This set is based on the location of local maxima and minima in the systolic rise of the pulse wave after fractional differintegration of the signal. Several optimization criteria have been proposed and applied to high-quality recordings of a database with subjects who were breathing at different rates while sitting or standing. When a proper fractional differintegration order is selected by using the RR time series as a reference, the agreement between the obtained IBI and RR is better than that for other state-of-the-art fiducial points. This work tested seven different traditional fiducial points. For the agreement analysis, the median standard deviation of the difference between the IBI and RR time series is 5.72 ms for the proposed fiducial point versus 6.20 ms for the best-performing traditional fiducial point, although it can reach as high as 9.93 ms for another traditional fiducial point. Other optimization criteria aim to reduce the standard deviation of the PAT (7.21 ms using the proposed fiducial point versus 8.22 ms to 15.4 ms for the best- and worst-performing traditional fiducial points) or to minimize the standard deviation of the PAT attributable to breathing (3.44 ms using the proposed fiducial point versus 4.40 ms to 5.12 ms for best- and worst-performing traditional fiducial points). The use of these fiducial points may help to better quantify the beat-to-beat PAT variability and IBI time series.

Reliability of industrial policies in Nepal: An empirical investigation into the role of macroeconomic indicators

This study aims to analyze the reliability of Nepal’s industrial policies, focusing on the effects of macroeconomic variables on implementation and outcomes. This paper assesses Nepal’s industrial policies, emphasizing the need for improvements, export promotion, and human capital development while recognizing the importance of strategic planning and context-specific approaches for economic growth, stability, and development. The analytical and descriptive approaches have been applied to analyze the data by collecting secondary data sources that include official publications, which encompass 47 time series variables from 1974 to 2020. The findings provide mixed evidence for the economic impacts of liberalization, with exports and liberalization driving overall GDP growth. In contrast, other factors like economic openness, tourism, and their relationship with industrial GDP remain statistically insignificant. The paper indicates that remittances and investment have the most substantial impact on GDP, raising it by 1.86 and 1.21 units per unit increase, respectively. Exports have a moderate impact on industrial GDP (0.403 units). Export-oriented industries and tourism lack significant associations with either type of GDP. Liberalization significantly boosted both GDP and industrial GDP, with an increase of 179465.3 and 49595.62 units, respectively. Imports also jumped post-liberalization, driven by higher remittances as 1.215 units per unit increase. This study on industrial policies in developing economies, focusing on Nepal, adds valuable insights. The findings can ensure policymaking, boost economic growth, and strengthen Nepal’s industrial sector.

A Comparison of the Rank-based and Slope-based Nonparametric Tests for Trend Detection in Climate Time Series

Trend detection in climate time series data is crucial for understanding climate change, predicting future climate patterns, assessing impacts, managing resources, and formulating policies. Several trend detection methods have been introduced in the literature, including parametric and non-parametric approaches. Nonparametric trend detection methods are often considered more preferable than parametric methods in certain situations due to their flexibility and robustness. Comparing various nonparametric methods of trend detection is vital in data analysis because different techniques can yield divergent results based on the same dataset. In this study, three nonparametric trend tests which were the MannKendall (MK), Sen’s Innovative Trend Analysis (ITA) and Modified Mann-Kendall by Sen’s Innovative Trend Analysis (MMK_ITA) were compared based on their power. The MK test is a rank-based test and the ITA is a slope-based test. Meanwhile, the combination of rank-based and slope-based methods is known as the MMK_ITA test. The power analysis was conducted through Monte Carlo simulation on normal, non-normal and autocorrelated time series. The simulation results indicated that test power relied on magnitude of linear trend slope, sample sizes, distribution type and variation in time series. These tests were then applied to monthly maximum temperature from 2002 until 2021 for Selangor, Malaysia. This study found that the slope-based test performed better compared to the rank-based test and their combined methods from the simulation studies and real data application based on the calculated power.