Time Series Of Variables Research Articles

The Spatiotemporal and Dependency Analysis of selected Meteorological Parameters and Normalized Difference Vegetation Index with Aerosol Optical Depth over East Africa.

The unprecedented rise in atmospheric aerosols, coupled with their intricate interactions with the environment through a wide array of physical, chemical, and biological processes, has profoundly impacted global climate. Their presence in the atmosphere scatters and absorbs solar radiation, thus altering the amount of sunlight reaching the Earth's surface. These direct effects, along with the indirect effects of aerosols, have significantly altered atmospheric temperatures, land surface processes, global surface temperature, hydrological cycle, and ecosystems. Understanding the complex interplay between aerosols and climatic variables necessitates a multidisciplinary approach, such as dependency modeling. Addressing these challenges, the current study conducts a spatiotemporal correlational analysis of selected key meteorological parameters with aerosol optical depth over East Africa (EA) using multisensory data from Moderate-resolution Imaging Spectroradiometer (MODIS), Modern-Era Retrospective analysis for Research and Application (MERRA-2) model, and Tropical Rainfall Measurement Mission (TRMM). Employing a weighted least squares regression (WLS) model, the study quantifies trends in the time series of climatic variables and Normalized Difference Vegetation Index (NDVI), further utilizing a statistical dependency modeling technique for correlational analysis. The trend analysis reveals a significant decreasing trend in surface wind speed (SWS) in most months, with sporadic positive trends attributed to anthropogenic activities, notably biomass burning, observed in January. Spatial trend analysis of Precipitation Rate (PR) displays heterogeneity, with significant negative trends in January and March, and positive trends in February, April, November, and December. Negative trends during May to August are attributed to increased anthropogenic activities, while enhanced positive trends in May correlate with low aerosol optical depth (AOD) during this period. Surface air temperature (SAT) exhibits diverse variations across the region, with dry months recording higher averages and trends than wet months. The study notes heterogeneous correlations in NDVI over the study area, with positive and negative correlations observed in different regions. Specifically, positive correlations are noted along the coastal and Lake Victoria regions, attributed to improved PR enhancing vegetation cover in these areas.

Improving cross-regional model transfer performance in crop classification by crop time series correction

ABSTRACT Cross-Regional Model Transfer (CRMT) provides a solution to crop classification challenges in target regions with limited labeled samples. However, when the source region (source domain) and the target region (target domain) are spatially distant, a shift in crop Satellite Image Time Series (SITS) occurs, resulting in suboptimal performance of SITS-based classification models in the target domain. To address this issue, we propose a novel domain adaptation method, named SITS Correction. This method mitigates the SITS shift by introducing correction factors to improve the classification performance of CRMT. The correction factors consist of source domain correction units and target domain correction units, which are from variables associated with the SITS shift. In cases where labeled samples are available in the target domain, the SITS have calculated Barycenter (i.e. SitsB) as the correction units, which then combine the source domain and the target domain’s SitsB to form a SITS-based correction factor. When no labeled samples are available, the correction units are substituted by Barycenter of the temperature variables time series (i.e. temperature indicator) influencing crop growth, and the correction factors are composed of temperature indicators from both the source and target domains. We validate SITS Correction through transfer experiments conducted in four regions, encompassing various years, crops, and classification methods. Experimental results demonstrate a significant improvement in crop classification performance with SITS Correction in both scenarios – target domain with labeled samples (accuracy increased by 39.1%) and target domain without labeled samples (accuracy increased by 14.8%). This research provides a practical option for crop classification in target regions with either limited or no labeled samples.

Correlation dimension for temperature and other micrometeorological variables in different ground cover from six biomes of the Amazon basin

Correlation dimension () for embedding dimensions, varying between 3 to 11, were estimated for a 1- or 2-month time series of air temperature and other micrometeorological variables, such as energy fluxes, solar radiation and concentration, in six ecosystems in Amazon and the central region of Brazil. Dc remained within the 1.5 - 3.0 range, for variables locally coupled to ecosystems, and outside the range for variables not locally coupled, such as precipitation and wind velocity. Results indicate that the feature of the variables does not have an important stochastic component. There is a small amount of data available concerning estimates of correlation dimension () of micrometeorological variables. The measurement of for several variables from six Brazilian biomes indicates that this parameter is sensitive to environmental conditions as the rainfall. The obtained results also indicates that the natural processes in the ecosystems remain in a non-chaotic regime. In certain circumstances, as in conditions of high humidity, natural ecosystems may stay inside a limit which is close but lower than chaos. Moreover, an important stochastic component was not observed.

Forecasting Non-stationary Time Series Using Deep Learning in a Fuzzy Time Series Framework and its Application to Stock Markets

Non-stationary time series prediction is challenging due to its dynamic and complex nature. Fuzzy time series models offer a promising solution for forecasting such data, but a key challenge lies in partitioning the universe of discourse, which significantly impacts forecasting accuracy. Traditional fuzzy time series models often use equal-length interval partitioning, which is more suited for stationary data and limits their adaptability to non-stationary time series. This paper introduces a novel variable-length interval partitioning method designed specifically for non-stationary time series. The developed method combines a Long Short-Term Memory (LSTM) Autoencoder with K-means clustering, enabling dynamic, data-driven partitioning that adapts to the changing characteristics of the data. The LSTM Autoencoder encodes the time series, which is clustered using K-means, and intervals are defined based on cluster centers. Furthermore, the Variable Length Interval Partitioning-based Fuzzy Time Series model (VLIFTS) is developed by incorporating this partitioning method and the concepts of Markov chain and transition probability matrix. In this model, fuzzy sets are viewed as states of a Markov chain, and transition probabilities are used in the forecasting phase. The model is validated on stock market indices Nifty 50, NASDAQ, S&P 500, and Dow Jones. Stationarity and heteroscedasticity are tested using Augmented Dickey-Fuller (ADF) and Levene's tests respectively. Statistical forecast accuracy metrics Root Mean Squared Error (RMSE) and Mean Absolute Percent Error (MAPE) show that VLIFTS significantly improves forecasting accuracy over traditional models. This hybrid approach enhances fuzzy time series modelling and can be applied to various non-stationary time series forecasting problems.

Hospital stays and probable dementia as predictors of relocation to long-term care facilities.

This study aims to investigate the relocation of older adults in the US from community living to long-term care facilities (LTCFs). Specifically, it examines the predictive roles of possible and probable dementia and hospital stays in this complex health care transition. Utilizing data from the National Health and Aging Trends Study, a longitudinal cohort study (2011-2019), we employed a panel data approach, which consists of multiple observations over time for the same participants, allowing us to account for both cross-sectional variations (differences between participants) and time-series variations (changes in the same participant over time). The analysis involved longitudinal logistic regression models. Using the AD8 dementia screening interview, clock drawing test, immediate and delayed word recall test, orientation, and history of dementia diagnosis, we placed participants into categories of having no dementia, possible dementia, and probable dementia. A survey asked about hospital stays in the past year. Relocation to LTCFs was examined based on the changes to the living location. The proportion of individuals transitioning to LTCFs tripled between 2011 and 2019, emphasizing the need to understand and manage this health care transition. Hospital stays significantly increased the probability of moving to LTCFs, especially nursing homes. Probable dementia demonstrated a 3-fold increase, aligning with the rising prevalence of Alzheimer disease. Difficulty walking and climbing stairs significantly increased relocation probabilities. The study findings emphasize complexity in late-life relocations influenced by dementia and hospital stays. Screening for cognitive function among community-dwelling older adults, particularly those with a history of hospital stays and mobility difficulties, can inform interventions and policies. Implications extend to health care policy, geriatric care, and the imperative for targeted interventions considering demographic variations. Future research should explore additional variables and address limitations to refine our understanding of the relocation process.

Effects of the kinematic variable, time delay and data length on test-retest reliability of the maximal Lyapunov exponent of human walking.

The maximal Lyapunov exponent (MLE) has been used to quantify the dynamic stability of human locomotion. The method for estimating MLE requires selecting a proper time series of kinematic variables and reconstructing phase space using proper time delay. The data length also affects the reliability of the measured MLE. However, there has been no criterion for the choice of the time series, time delay or data length. Here, we quantified the effect of these factors on the test-retest reliability of MLE estimations. We recruited 15 young and healthy adults and let them walk on a treadmill three times. We calculated MLE employing various lengths of time series of 18 frequently used kinematic variables and two typical choices of time delay: fixed delay and delay selected by average mutual information algorithm. Then, we measured the intraclass correlation coefficient (ICC) of the measured MLE under each condition. Our results show that the choice of time delay does not affect reliability. Five among the 18 kinematic variables enabled excellent reliability with ICC above 0.9 within 450 strides and also enabled ICC above 0.75 even with 60 or less strides. These findings can contribute to establishing the criteria for measuring the dynamic stability of human walking.

Gated recurrent units for modelling time series of soil temperature and moisture: An assessment of performance and process reflectivity

Soil temperature and moisture are important variables controlling ecological processes, but continuous high-resolution data are rarely available. Therefore, we used the correlation with widely accessible meteorological variables, including air temperature and precipitation, to develop models that predict time series of soil temperature and moisture. To model high-resolution time series, predictor and target variables had a temporal resolution of 1 h. We tested the applicability of Gated Recurrent Units with time series from one exemplary site. The models showed a high predictability on the four years test set with a mean absolute error of 0.87 °C for soil temperature and 3.20% volumetric water content for soil moisture. We further investigated the plausibility of the models by passing simplified synthetic data to the trained models and thereby proved their ability to reflect known processes. Finally, we showed the potential to apply the models to other sites and soil depths using transfer learning.

Secure smart grid implementation with automatic data integrity attack location prediction and exalted energy theft detection

Smart grid systems use digital communication technology to efficiently monitor and manage electricity distribution. However, there is a need for improve the robustness of smart grid systems. Hence a novel “Secure Smart Grid Implementation with Automatic Data Integrity Attack Location Prediction and Exalted Energy Theft Detection” has been proposed. Current detection methods are not good at identifying data integrity attacks. So a novel Automatic Deterministic Attack Location Prediction Model detects data integrity attacks using a Stacked Watermarking BiGRU-based Deep Deterministic Q network, extracting system matrix and state time series variation. Moreover, Energy theft detection algorithms, which are intended to detect dishonest clients, provide erroneous findings because of linear interpolation processes and randomised near misses, which ignore important data points. Thus a novel Exalted Energy Theft Detector with Secure Encoding is introduced where Amortized Multicode GAN method secures smart grids by detecting energy theft and preventing firmware code modifications, utilising fractional multi-level quantum encoding to identify injection locations. As a result, the proposed model has an accuracy of 97.1%, recall of 96.25%, error rate is 1.36%, F1-score is 96.20%, sensitivity is 96.10%, specificity is 96.10%, MAPE is 0.48%, MSE is 0.0001 and RMSE is 0.017.

Evaluation of the space-time effects of Covid-19 on household loans and savings in Romania - A spatial panel data approach at county level

This article examines the effects of the Covid-19 crisis on the banking sector of Romania using a spatial panel dataset of 21 months, from April 2020 to December 2021, among 41 counties. The empirical setup exploits the spatial and time-series variations of the banking sector during the pandemic via means of exploratory data analysis, simple linear models for each month with spatial diagnosis, and spatial panel data models with fixed effects. County-level differences and particular disturbances in the trend of banking services are highlighted against the turning points of the pandemic. We show that increasing Covid-19 infections result in higher loans and savings per capita, with significant spatial interactions given by the neighboring counties.

Motor variability during resistance training: Acceleration signal as intensity indicator

Analysis of variability in physiological time series has been shown to be an indicator of the state of the organism. Although there is evidence of the usefulness of analysis of the amount and/or structure of variability (complexity) in cycling actions, there is limited knowledge about its application in resistance exercise. The aim of this study is to find out whether variability in acceleration signals can be an indicator of intensity level in a squat task. For this purpose, an experimental design was developed in which the following participated seventy-two participants (age = 25.7 ± 4.4 years; height = 169.2 ± 9.8 cm; body mass = 67.7 ± 11.2 kg; ratio 1RM/body mass = 1.4 ± 0.3). They performed four repetitions of back squat at loads of 10%, 30%, 50%, 70%, and 90% of 1RM. Acceleration during the exercise was recorded using an inertial measurement unit (IMU) and a force platform. The variability of the movement was then analyzed using Standard Deviation (SD), Detrended Fluctuation Analysis (DFA), Fuzzy Entropy (FuzzyEn), and Sample Entropy (SampEn). For the IMU and for the force platform, significant effects were observed in all variables (p < 0.001). In pairwise comparisons, IMU showed a significant increase in motor complexity with increasing intensity, among most intensities, in DFA, FuzzyEn and SampEn. Differences in force platform were more limited, and only DFA detected differences between most intensities. The results suggest that measures of signal and acceleration variability may be a useful indicator of the relative intensity at which a squat exercise is performed.

Assessing personal PM2.5 exposure using a novel neck-mounted monitoring device in rural Rwanda

There is growing global concern regarding the detrimental health impacts of PM2.5 emissions from traditional stoves that utilize polluting fuels. Conventional methods for estimating daily personal PM2.5 exposure involve personal air samplers and measuring devices placed in a waist pouch, but these instruments are cumbersome and inconvenient. To address this issue, we developed a novel neck-mounted PM2.5 monitoring device (Pocket PM2.5 Logger) that is compact, lightweight, and can operate continuously for 1 week without recharging. Twelve participants who utilized charcoal, firewood, or propane gas for cooking in rural regions of Rwanda wore the Pocket PM2.5 Logger continuously for 1 week, and time-series variations in personal PM2.5 exposure were recorded at 5-min intervals. Individual daily exposure concentrations during cooking differed significantly among users of the different fuel types, and PM2.5 exposure was at least 2.6 and 3.4 times higher for charcoal and firewood users, respectively, than for propane gas users. Therefore, switching from biomass fuels to propane gas would reduce daily individual exposure by at least one-third. An analysis of cooking times showed that the median cooking time per meal was 30 min; however, half the participants cooked for 1.5 h per meal, and one-third cooked for over 4.5 h per meal. Reducing these extremely long cooking times would reduce exposure with all fuel types. The Pocket PM2.5 Logger facilitates the comprehensive assessment of personal PM2.5 exposure dynamics and is beneficial for the development of intervention strategies targeting household air pollution.

Multiperiod Dynamic Portfolio Choice: When High Dimensionality Meets Return Predictability

Multiperiod portfolio choice is the central problem in active asset management. Multiperiod dynamic portfolios are notoriously difficult to solve, especially when there are hundreds of tradable assets as well as a large number of state variables. In this article, we develop a novel two-step methodology to solve the multiperiod dynamic portfolio choice problem with high dimensional assets in the presence of return predictability conditional on a large number of state predictors. Specifically, in the first step, we propose the new Risk-Premium Projected-PCA (RP-PPCA) method to reduce the dimension of tradable assets. This method achieves Dimension Reduction (DR) by estimating latent factors with explanatory power in both time series variation and expected return in high-dimension-low-sample-size data. In the second step, we use dynamic programming to solve the multiperiod portfolio choice problem, and in each recursive step, we adopt an Adjusted semiparametric Model Averaging (AMA) method to avoid the curse of dimensionality associated with a large set of state variables while remaining computationally efficient. Thus, we name this two-step approach DRAMA, which stands for a combination of a new dimension reduction method and an adjusted semiparametric model averaging method. Analytically, we show that the portfolios constructed by the DRAMA are approximately optimal under mild assumptions. Moreover, our numerical results based on empirical data from US stock markets show that the proposed portfolios have excellent out-of-sample performances.

Impact of higher education on economic growth: A vector error correction analysis in malawi

This study examined the impact of higher education on economic growth in Malawi from 1995 to 2021. The study contributes to existing literature by providing more recent evidence and as well contributes to the debate on whether higher education has a positive or negative impact on economic growth in a developing country like Malawi. Empirical studies have so far yielded inconclusive results. The time series variables were tested to find out its order of integration using Dickey Fuller and the series were all non-stationary at level but became stationary after first difference. The study employed cointegration and evidence of cointegration of higher education and economic growth was established. However, Vector Error Correction Model estimates showed that there is negative long run relationship between higher education and economic growth. Thus, the government and other relevant education stakeholders should strive to increase enrollment in tertiary institutions so that it yields significant impact on economic output.

Multifractal Hopscotch in Hopscotch by Julio Cortázar.

Punctuation is the main factor introducing correlations in natural language written texts and it crucially impacts their overall effectiveness, expressiveness, and readability. Punctuation marks at the end of sentences are of particular importance as their distribution can determine various complexity features of written natural language. Here, the sentence length variability (SLV) time series representing Hopscotch by Julio Cortázar are subjected to quantitative analysis with an attempt to identify their distribution type, long-memory effects, and potential multiscale patterns. The analyzed novel is an important and innovative piece of literature whose essential property is freedom of movement between its building blocks given to a reader by the author. The statistical consequences of this freedom are closely investigated in both the original, Spanish version of the novel, and its translations into English and Polish. Clear evidence of rich multifractality in the SLV dynamics, with a left-sided asymmetry, however, is observed in all three language versions as well as in the versions with differently ordered chapters.

Application of Vector Autoregressive (VAR) Model on the Interaction of Inflation Rates and Public Debt in Kenya from 2011 to 2021

This study examines the relationship between public debt and inflation rates in Kenya from 2011 to 2021 using the Vector Autoregressive (VAR) model. Despite the models likeAutoregressive Integrated Moving Average (ARIMA), Seasonal Autoregressive Integrated Moving Average (SARIMA), and Generalized Autoregressive Conditional Heteroscedasticity (GARCH) gaining popularity in time series analysis, the Vector Autoregressive model, being multivariate, is relevant in analyzing two or more time series variables simultaneously, benefiting from the bi-directional causality and providing a better outlook into the flow of the dynamic interaction between inflation and public debt. The main objectives are modelling the Vector Autoregressive model and forecasting future trends to provide insights for policymakers. Additionally, the methodological approach comprises descriptive statistics, stationarity tests, normality tests, and the Vector Autoregressive model. Descriptive statistics reveal significant variations, with public debt increasing from 1.35 trillion KES to a peak of 8.2 trillion KES, and inflation rates ranging from 3.2% to 19.72% for the period from 2011 to 2021. The Augmented Dickey-Fuller (ADF) test confirmed that both time series were stationary at their levels. The Vector Autoregressive model, chosen for its ability to analyze dynamic interactions, indicated a significant relationship between the variables, with inflation showing strong self-persistence (coefficient of 0.8731, <i>p</i> < 2 × 10<sup>−16</sup>), though public debt did not significantly impact inflation in the model (<i>p</i> = 0.5592). The models R-squared values, 95.82% for public debt and 84.74% for inflation, highlight its strong explanatory power. Moreover, findings indicate that while public debt does not directly affect inflation within the model lag structure, inflation exhibits a strong self-persistence. The model R-squared values are 95.82% for public debt and 84.74% for inflation, demonstrating high explanatory power. Recommendations include the implementation of a robust debt management strategy, emphasizing sustainable borrowing and enhancing revenue generation to mitigate inflationary pressures. Further research is recommended to explore the broader macroeconomic impacts of public debt on economic growth and employment in Kenya.

Designing an explainable bio-inspired model for suspended sediment load estimation: eXtreme Gradient Boosting coupled with Marine Predators Algorithm

This study aimed to develop an accurate and reliable model for predicting suspended sediment load (SL) in river systems, which is crucial for water resource management and environmental protection. While Xtreme Gradient Boosting (XGB), a powerful ensemble machine learning (ML) model, has been employed in previous studies, the novelty of this research lies in the introduction of a hybrid approach that synergistically combines XGB with the bio-inspired Marine Predators Algorithm (XGB-MPA) to estimate SL in the Yeşilirmak River (Turkey). To this end, streamflow (Q) and sediment concentration (SC) values as well as their lag times (1 to 3 month lag times) were fed as input variables – under 9 scenarios – into ML models. A time series of datasets from March 1973 to December 2011 and January 2012 to March 2023 were used for training and testing of ML models, respectively. The superiority of the proposed model (XGB-MPA) compared to two other hybrid models, including XGB-PSO (Particle Swarm Optimization) and XGB-GWO (Grey Wolf Optimization) was also investigated. According to the results, the simultaneous application of Q and SC lag time values as inputs has led to the best SL estimates by XGB-MPA, with XGB-MPA9 (RMSE = 103.7 ton/day; NSE = 0.96) exhibiting the lowest error rates. In addition, XGB-MPA has performed better than XGB in all scenarios, with the lowest and highest reduction in RMSE being 19.3% (scenario 5) and 97.4% (scenario 1), respectively. When comparing the performance of hybrid models, the proposed XGB-MPA model has performed best with MAE, RMSE and NSE of 40.94, 103.7 and 0.96, respectively, in comparison with 816.02, 1063.74 and −2.94 for XGB-PSO and 693.16, 981.68 and −2.37 for XGB-GWO. Further research can include the use of time series of efficient variables extracted from satellite images (e.g. land cover, river morphology, etc.) as model inputs. Highlights Improvement of SL estimation by coupling MPA with XGB model Using delayed combinations of streamflow and sediment concentration as model inputs Superiority of MPA compare to PSO and GWO in SL estimation Greater variations in SHAP caused by sediment concentration compared to streamflow Further studies required on the effects of hydrological and topographical features

Comparing the hedge and safe haven properties of individual commodities for China and United States equity sectors

PurposeThe purpose of this paper is to scrutinize the safe haven benefits of 13 individual commodities for the USA and Chinese equity sectors during the financial turmoil period. Therefore, sectoral investors in the USA and China could invest in those specific commodities that provide stable returns during the health crisis and financial turmoil periods.Design/methodology/approachThe daily data spans from February 1, 2015, to July 28, 2022. The present study applies several different approaches to analyzing the data set. The author apply the cross-quantilogram (C.Q) methodology to capture the lead-lag bivariate quantile interdependence between two stationary time series variables during the bearish, bullish and normal periods. Then the study used the hedging effectiveness (HE) and conditional diversification benefits (CDB) approaches to capture the hedging and diversification benefits of commodity classes and individual commodities.FindingsThe noteworthy findings of the quantilogram methodology reveal that livestock and agriculture commodities serve as better refuges as compared to the precious metals and energy index in both countries. On average, precious metals failed to serve as safe haven investments for the USA and Chinese equity market sectors. All energy commodities except soybean oil had strong comovements with China and the US equity sectors during bearish, bullish and normal periods. Lean hogs, fiddler cattle and live cattle are perfect hedging assets for both countries due to the presence of blue color at normal and bullish periods in all C.Q heat-maps. The HE table depicts that commodity indices and individual commodities failed to serve as hedging assets for the Chinese equity sectors. But commodities are semistrong hedging assets for the US equity sectors and the S&P 500 due to the average HE values being 0.7 and above. The CDB values depict that precious metals provide diversification benefits in both equity markets.Practical implicationsThe present study results have important implications for equity sector investors of the USA and China in suggesting particular commodity during the financial turmoil period. During the bearish market condition, risk averse equity sector investors can invest in livestock commodities and agriculture commodities, due to their relatively stable returns. In addition, policymakers can use the analysis insights to formulate policy tools and monitoring mechanisms, effectively mitigating the unfavorable effects arising from asymmetric dependence between commodities and equity sectors during the upper tail, middle and lower tail. Policymakers can suggest equity investors to invest in which commodity during extreme conditions.Originality/valueThe current study has the following points of originality. First, to the best of the author’s knowledge, this is the first study to investigate the individual commodities’ roles as safe havens taken from all four major commodity classes. More importantly, it is also noticeable that the safe haven abilities of commodities are usually tested for the stock market, but the equity sectors are ignored. Therefore, the present study used both stock market and sectoral indices data.

Advancing Process-Oriented Geographical Regionalization Model

Existing regionalization methods have largely overlooked the temporal dimension, leading to outcomes that predominantly reflect spatial differentiation of regional variables only at a singular temporal instance, rather than across their entire evolutionary process. In response, this research proposes a novel model for geographic process regionalization based on spatial time-series (STS) data. This model is designed to amalgamate adjacent entities exhibiting analogous trends in regional variable fluctuations, while segregating those with disparate trajectories. The study starts by establishing methodologies for the assessment of three distinct types of STS data similarities: value-based, shape-oriented, and periodic patterns. Following this, a tailored regionalization methodology specific to STS data is delineated. This methodology facilitates a quantitative analysis of time-series variables within each geographical region, leading to the development of a process-oriented geographic boundary model (GBM). The efficacy of this model is validated through empirical testing with synthetic data sets and further elucidated via case studies focusing on the dynamics of population and pollution levels, underscoring its substantial applicability in both human and physical geography. The proposed methodology is adaptable for any time-series data that incorporate locational information for regionalization.

Machine learning potentials for global multi-timescale diffuse irradiance estimation: Synthesizing ground observations, time-series, and environmental features

Separating diffuse horizontal irradiance (DHI) from ground-based global horizontal irradiance observations is critical owing to lack of direct DHI observations. Existing models are often site-specific and time-bound, thereby limiting their universal applicability. To address this, this study thoroughly explores machine learning (ML) for constructing global separation models. We develop 36 models using three ML algorithms—light gradient boosting machine (LightGBM), generalized additive model (GAM), and geographically weighted artificial neural network (GWANN)—alongside three data splitting strategies for minutely, 10-minutely, hourly, and daily timescales. LightGBM and GAM outperform GWANN in accuracy, with LightGBM excelling in interpretability, efficiency, and handling missing values, while GWANN exhibits superior stability. Compared with completely random splitting, model accuracy decreases by 2.26 % and 2.16 % for station- and date-based splitting, respectively. Prediction accuracy varies across timescales, with minutely models typically considered the best. Key predictors are Kt, solar zenith angle, and altitude, complemented by the significant influence of time-series Kt, aerosol, and cloud features. The influence of various factors on model accuracy differs and is scale-dependent. Among them, time-series Kt variability factors notably enhance ML-based predictions, especially at minutely scales. The LightGBM model outperforms classic models in overall and site-specific accuracy and adaptability, but its computational efficiency declines, especially with time-series variability factors. The findings highlight that ML-based separation models necessitate careful selection of algorithm, data splitting strategy, and input variables, fully considering study region, data situation, and temporal and spatial scales. This research offers a potential solution for constructing global separation models and insights for model improvement.

Student Motivation Analysis Based on Raising-Hand Videos.

In current smart classroom research, numerous studies focus on recognizing hand-raising, but few analyze the movements to interpret students' intentions. This limitation hinders teachers from utilizing this information to enhance the effectiveness of smart classroom teaching. Assistive teaching methods, including robotic and artificial intelligence teaching, require smart classroom systems to both recognize and thoroughly analyze hand-raising movements. This detailed analysis enables systems to provide targeted guidance based on students' hand-raising behavior. This study proposes a morphology-based analysis method to innovatively convert students' skeleton key point data into several one-dimensional time series. By analyzing these time series, this method offers a more detailed analysis of student hand-raising behavior, addressing the limitations of deep learning methods that cannot compare classroom hand-raising enthusiasm or establish a detailed database of such behavior. This method primarily utilizes a neural network to obtain students' skeleton estimation results, which are then converted into time series of several variables using the morphology-based analysis method. The YOLOX and HrNet models were employed to obtain the skeleton estimation results; YOLOX is an object detection model, while HrNet is a skeleton estimation model. This method successfully recognizes hand-raising actions and provides a detailed analysis of their speed and amplitude, effectively supplementing the coarse recognition capabilities of neural networks. The effectiveness of this method has been validated through experiments.