The global coffee market is a complex and dynamic system influenced by a multitude of factors, resulting in continuous fluctuations in coffee prices. This study utilized time series analysis to examine the historical trends and dynamics of the monthly global price of coffee, Arabica, from January 1990 to July 2023. The data were transformed to achieve stationarity using a methodical process, and an ARIMA (0,1,2) model was found to be the best-fit model for forecasting. The analysis' findings show how complex the coffee market is. Coffee prices are influenced by factors like supply and demand, climate change, currency exchange rates, economic conditions, and trade policies. Since 2001, unstable markets have affected producers and consumers. The COVID-19 pandemic brought unprecedented challenges to the coffee sector, disrupting consumption patterns and supply chains. Economic variables like GDP growth and exchange rates influence coffee prices and producer welfare. While the coffee industry is experiencing recovery, price fluctuations remain a concern. Understanding these factors is crucial for stakeholders, and time series analysis can help inform decision-making in this dynamic market.

This study tried to demonstrate the role of Time Series models in a modeling and forecasting process using publicly available long-term records of monthly global price of bananas during the period of January 1990 to November 2021 reported in the International Monetary Fund. Following the Box–Jenkins methodology, an ARIMA (2,1,4) with a drift model was selected as the best-fit model for the Time Series, according to its lowest AIC value. Using the Levenberg-Marquardt algorithm, the results revealed that the NARNN model with 12 neurons in the hidden layer and 6 times delays provided the best performance in the nonlinear autoregressive neural network models at its smaller MSE value. The ARIMA and NARNN models are good at modelling linear and nonlinear problems for the Time Series, respectively. However, using the HYBRID model, a combination of the ARIMA and NARNN models that has both linear and nonlinear modeling capabilities can be a better choice for modeling the Time Series. The comparative results revealed that the HYBRID model with 11 neurons in the hidden layer and 3 times delays yielded higher accuracy than the NARNN model with 12 neurons in the hidden layer and 6 times delays, and the ARIMA (2,1,4) with a drift model, according to its lowest MSE in this study.

The study employed data from the publicly accessible dataset titled "Marine and Coastal Ecosystems and Climate Change: Dataset from a Public Awareness Survey" to investigate public perceptions regarding significant dangers to oceanic and shoreline ecosystems. Survey participants assessed the severity of 13 notable threats, such as fishing, plastic pollution, climate change, shipping, and coastal development, using a psychometric rating scale of 1 (very low) to 5 (very high), with an additional option for "I don't know." The primary aim of the study was to classify respondents through principal component analysis (PCA)-based k-means clustering. This analysis identified three distinct participant clusters, each reflecting varying perspectives on the severity of these threats. This PCA-based k-means clustering offered crucial insights into the diverse ways different population segments perceive and prioritize environmental risks. By mapping these perceptions, the findings provided valuable guidance for policymakers, emphasizing the necessity for customized, data-driven strategies in environmental conservation and marine protection. These insights can bolster efforts to align public concern with effective climate change mitigation and targeted interventions to safeguard marine ecosystems.

This study utilized a mixed model approach, incorporating k-means clustering analysis for data examination, discriminant analysis for classification, and multilayer perceptron neural network analysis for prediction. After removing inadequate samples and outliers, the total number of observations was 19,692 for this study, which was collected through an interactive online personality test (i.e., Big Five Personality Traits) in 2012. The empirical results based on the k-means clustering analysis identified four different personality clusters using the total score of Big Five Personality Traits (Extraversion, Neuroticism, Agreeableness, Conscientiousness, and Openness to Experience). The empirical results obtained from the k -means clustering analysis revealed the presence of four distinct personal clusters, determined by the total scores of the Big Five Personality Traits. The accuracy of the clustering analysis was further tested using discriminant analysis, which indicated significant difference among the cluster means and correctly classified 95.5% of the original grouped cases. For predictive modeling, a multilayer perceptron neural network framework was used. The network had a 5-6-4 structure and was employed to determine the personality classification of participants. Notably, the model achieved 99.4% accuracy in correctly classifying the training grouped cases and 99.2% accuracy for the testing grouped cases. The results of this study offer valuable insights into understanding the personalities of participants, with implications for various domains such as psychology, social sciences, cultural studies, and economics.

This study aimed to demonstrate the effectiveness of time series models in modeling long-term records of global average absolute sea level changes from 1880 to 2014. Following the Box–Jenkins methodology, the ARIMA(0,1,2) model with drift was identified as the best-fit model for the time series due to its lowest AIC value. Using the LM algorithm, the results revealed that the NARNN model with 7 neurons in the hidden layer and 7 time delays exhibited the best performance among the nonlinear autoregressive neural network models, as indicated by its lower MSE. While ARIMA models excel in modeling linear problems within time series data, NARNN models are better suited for nonlinear patterns. However, a HYBRID model was explored, which combines the strengths of both ARIMA and NARNN models, offering the capability to address both linear and nonlinear aspects of time series data. The comparative analysis of this study demonstrated that the HYBRID model, with 6 neurons in the hidden layer and 7 time delays, outperformed the NARNN model with 7 neurons in the hidden layer and 7 time delays, as well as the ARIMA(0,1,2) model, with the lowest MSE in this study. These findings represent a significant step in time series forecasting by leveraging the strengths of both statistical and machine learning methods.

Accurate forecasting of mean sea level (MSL) is critical for coastal risk management, infrastructure planning, and climate adaptation, particularly along the low-lying Mid-Atlantic coast of the United States. This study presents a comprehensive comparative evaluation of four data-driven deep learning and neural network architectures, including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), one-dimensional Convolutional Neural Network (CNN), and a lagged feed-forward neural network (FFNN), applied to seasonally adjusted monthly MSL records at the NOAA tide gauge station in Ocean City, Maryland (Station 8570283), spanning August 2002 through February 2025. All models share a standardized 24-month lookback window (lags 1–13 for the FFNN), 80/20 train/test split, Adam optimization, and early stopping, enabling controlled comparison. Models are evaluated using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R²) on held-out test data. Results indicate that the GRU achieves the best predictive performance among the neural architectures (RMSE = 0.0704 m, MAE = 0.0587 m, R² = 0.099), followed by the LSTM (RMSE = 0.0734 m, R² = 0.020). Even the GRU explained only about 10% of the test-period variance, and its RMSE was only about 3% lower than that of a persistence forecast. The CNN and FFNN both yield negative R² values (−0.219 and −0.907 respectively), which on this single test period placed the recurrent models ahead of the convolutional and feed-forward networks; this ranking was not stable across test periods (see below). A 24-month ensemble forecast is presented for all models through February 2027. The FFNN ensemble spread (±1σ across 20 independently initialized networks) is reported only as a measure of initialization variability; calibrated prediction intervals are derived from the ARIMA model and from rolling-origin forecast errors. When benchmarked under the identical protocol against persistence, seasonal-naïve, random walk with drift, linear trend, ARIMA, SARIMA and exponential smoothing (ETS) baselines, however, all three statistical time-series models outperformed every neural architecture: ARIMA(3,0,0) with a linear trend achieved the lowest test error (RMSE = 0.0642 m, R² = 0.250), and the GRU improved on a simple persistence forecast by only 3% in RMSE. These findings indicate that, for this univariate monthly record, the neural architectures tested did not add predictive value beyond well-specified linear time-series models; whether this holds at other stations has not yet been tested. An expandingwindow rolling-origin evaluation over 49 forecast origins (December 2012 – December 2024), with every model re-trained or re-estimated at each origin, confirmed this ranking at forecast horizons of 1, 3, 6, 12 and 24 months: ARIMA had the lowest or near-lowest RMSE at every horizon (0.056–0.069 m), whereas the best neural model was 6–15% less accurate, and the single-split ranking of the neural architectures was not stable across test periods. All models predict the seasonally adjusted monthly MSL level in meters; training the networks on residuals from a training-period trend improved their accuracy and stabilised their forecasts but did not change their ranking relative to the statistical models. The NNAR model of R’s nnetar(), added as a separate benchmark, was far more accurate than the FFNN on the single split (RMSE = 0.0708 m) but was unstable at multi-step horizons and did not outperform ARIMA. Backtesting of the recursive multi-step forecasts showed that the neural models increasingly under-forecast with lead time, and that ARIMA prediction intervals were close to nominal coverage, whereas the FFNN ensemble spread and the simulated nnetar() intervals were far too narrow to serve as prediction intervals.