Historical Dataset Research Articles

Digital Twins’ Advancements and Applications in Healthcare, Towards Precision Medicine

This review examines the significant influence of Digital Twins (DTs) and their variant, Digital Human Twins (DHTs), on the healthcare field. DTs represent virtual replicas that encapsulate both medical and physiological characteristics—such as tissues, organs, and biokinetic data—of patients. These virtual models facilitate a deeper understanding of disease progression and enhance the customization and optimization of treatment plans by modeling complex interactions between genetic factors and environmental influences. By establishing dynamic, bidirectional connections between the DTs of physical objects and their digital counterparts, these technologies enable real-time data exchange, thereby transforming electronic health records. Leveraging the increasing availability of extensive historical datasets from clinical trials and real-world sources, AI models can now generate comprehensive predictions of future health outcomes for specific patients in the form of AI-generated DTs. Such models can also offer insights into potential diagnoses, disease progression, and treatment responses. This remarkable progression in healthcare paves the way for precision medicine and personalized health, allowing for high-level individualized medical interventions and therapies. However, the integration of DTs into healthcare faces several challenges, including data security, accessibility, bias, and quality. Addressing these obstacles is crucial to realizing the full potential of DHTs, heralding a new era of personalized, precise, and accurate medicine.

Research on passenger flow control at metro transfer stations based on real-time flow calculation of streamlines

PurposeThe volume of passenger traffic at metro transfer stations serves as a pivotal metric for the orchestration of crowd flow management. Given the intricacies of crowd dynamics within these stations and the recurrent instances of substantial passenger influxes, a methodology predicated on stochastic processes and the principle of user equilibrium is introduced to facilitate real-time traffic flow estimation within transfer station streamlines.Design/methodology/approachThe synthesis of stochastic process theory with streamline analysis engenders a probabilistic model of intra-station pedestrian traffic dynamics. Leveraging real-time passenger flow data procured from monitoring systems within the transfer station, a gradient descent optimization technique is employed to minimize the cost function, thereby deducing the dynamic distribution of categorized passenger flows. Subsequently, adhering to the tenets of user equilibrium, the Frank–Wolfe algorithm is implemented to allocate the intra-station categorized passenger flows across various streamlines, ascertaining the traffic volume for each.FindingsUtilizing the Xiaozhai Station of the Xi’an Metro as a case study, the Anylogic simulation software is engaged to emulate the intra-station crowd dynamics, thereby substantiating the efficacy of the proposed passenger flow estimation model. The derived solutions are instrumental in formulating a crowd control strategy for Xiaozhai Station during the peak interval from 17:30 to 18:00 on a designated day, yielding crowd management interventions that offer insights for the orchestration of passenger flow and operational governance within metro stations.Originality/valueThe construction of an estimation methodology for the real-time streamline traffic flow augments the model’s dataset, supplanting estimated values derived from surveys or historical datasets with real-time computed traffic data, thereby enhancing the precision and immediacy of crowd flow management within metro stations.

Trend‐Based Predictive Maintenance and Fault Detection Analytics for Photovoltaic Power Plants

Optimized predictive maintenance in photovoltaic (PV) systems is crucial for ensuring prolonged operational performance and cost‐effective operation and maintenance (O&M). Even though failure detection methods have already been developed, the main challenge remains the lack of predictive maintenance strategies to accurately forecast underperformance conditions. The scope of this work is to develop a predictive maintenance and failure detection routine for assessing the health status of PV systems. The workflow consists of the eXtreme gradient boosting algorithm for modeling the PV performance, the one‐class support vector machine algorithm for fault detection, and the Facebook Prophet algorithm for forecasting PV performance trends and generating maintenance alerts. The developed data‐driven routine analyzes performance trend deviations and it is validated using a historical dataset from a utility‐scale PV power plant in Greece. The obtained results show the effectiveness of the developed workflow in detecting fault conditions, achieving a sensitivity of 96.9%. Additionally, the results demonstrate the workflow's ability to generate predictive maintenance alerts up to 7 days in advance, yielding a sensitivity of 92.9%. Finally, the study provides useful insights that enhance operators’ efficiency in conducting O&M activities.

Petri net modeling and analysis of an IoT‐enabled system for real‐time monitoring of eggplants

AbstractAs the current agricultural industry is facing several challenges such as climate changes and lack of qualified farmers, it is extremely necessary to ensure sustainable agriculture and food supply by smart farming (SF). The SF can assist farmers and the associated stakeholders in making correct decisions on improving the yield and quality of agricultural products. In this study, an SF system based on low‐cost Raspberry Pi and Internet of Things (IoT) technologies has been successfully implemented. Through the data items detected by sensors to deeply manage the planting process, the IoT‐enabled communication protocol under ISO standards of Message Queuing Telemetry Transport (MQTT) was used. The bar charts of real‐time environmental parameters are presented on ThingsBoard to achieve the goal of data visualization. Meanwhile, a web server is built for the customized requirements, and the historical datasets are stored in the SQLite database system. Furthermore, a Petri net (PN) model was employed to detect all possible abnormal processes and to verify the feasibility and soundness of an IoT‐enabled system by using a software tool, WoPeD. Finally, the experimental results show that the plants cultivated by the proposed IoT‐enabled system are superior to those cultivated by the existing systems in many perspectives, including the monitoring distance, power consumption, precision, deadlock detection, and running time.

A genome-based survey of invasive pneumococci in Norway over four decades reveals lineage-specific responses to vaccination

BackgroundStreptococcus pneumoniae is a major cause of mortality globally. The introduction of pneumococcal conjugate vaccines (PCVs) has reduced the incidence of the targeted serotypes significantly, but expansion of non-targeted serotypes, serotype replacement, and incomplete vaccine-targeting contribute to pneumococcal disease in the vaccine era. Here, we characterize the changing population genetic landscape of S. pneumoniae in Norway over a 41-year period (1982–2022).MethodsSince 2018, all cases of invasive pneumococcal disease have undergone whole-genome sequencing (WGS) at the Norwegian Institute of Public Health. In order to characterize the changing population over time, historical isolates were re-cultured and sequenced, resulting in a historical WGS dataset. Isolates were assigned to global pneumococcal sequence clusters (GPSCs) using PathogenWatch and assigned to serotypes using in silico (SeroBA) and in vitro methods (Quellung reaction). Temporal phylogenetic analyses were performed on GPSCs of particular interest.ResultsThe availability of WGS data allowed us to study capsular variation at the level of individual lineages. We detect highly divergent fates for different GPSCs following the introduction of PCVs. For two out of eight major GPSCs, we identified multiple instances of serotype switching from vaccine types to non-vaccine types. Dating analyses suggest that most instances of serotype switching predated the introduction of PCVs, but expansion occurred after their introduction. Furthermore, selection for penicillin non-susceptibility was not a driving force for the changing serotype distribution within the GPSCs over time.ConclusionsPCVs have been major shapers of the Norwegian disease-causing pneumococcal population, both at the level of serotype distributions and the underlying lineage dynamics. Overall, the introduction of PCVs has reduced the incidence of invasive disease. However, some GPSCs initially dominated by vaccine types escaped the effect of vaccination through expansion of non-vaccine serotypes. Close monitoring of circulating lineages and serotypes will be key for ensuring optimal vaccination coverage going forward.

Scenario Generation Based on Ant Colony Optimization for Modelling Stochastic Variables in Power Systems

Uncertainty is an important subject in optimization problems due to the unpredictable nature of real variables in the power system area, which can condition the solution’s accuracy. The effective modelling of stochastic variables can contribute to the reduction in losses in the system under evaluation and facilitate the implementation of an effective response in advance. To model uncertainty variables, the most extended technique is the scenario generation (SG) method. This method evaluates possible combinations of complete curves. Classical scenario generation methods are founded on probability distributions or robust stochastic optimization. This paper proposes a novel approach for constructing scenarios using the Ant Colony Optimization (ACO) algorithm, referred to as ACO-SG. This methodology does not require a previous statistical study of uncertainty data to generate new scenarios. A historical dataset and the desired number of scenarios are the inputs inserted into the algorithm. In the case study, the algorithm used historical data from the Savona Campus Smart Polygeneration Microgrid of the University of Genoa. The approach was applied to generate scenarios of photovoltaic generation and building consumption. The low values of the Euclidean distance were used in order to check the validity of the scenarios. Moreover, the error deviation of the scenarios generated with the goal of daily power were 1.77% and 0.144% for the cases of PV generation and building consumption, respectively. The different results for both cases are explained by the characteristics of the specific cases. Despite these different results, both were significantly low, which indicates the capability of the algorithm to generate any kind of feature within curves and its adaptability to any case of SG.

Calibration of a hybrid model for HVAC systems for fault data generation

The application of automated fault detection and diagnosis algorithms has proven to be an effective way to increase the energy efficiency of buildings. While data-driven strategies have demonstrated their potential in developing such algorithms, they require a set of historical data that represents both healthy and faulty equipment operation, which is not a common scenario in real facilities. This article presents a methodology for the generation of synthetic data on both healthy and faulty operation for building heating and air conditioning equipment, which is especially useful for supervision and failure detection in the case of replication and expansion of buildings in tertiary or residential building planning programmes. This work investigates the automatic calibration of a typical air-handling unit hybrid model, composed of the detailed parametric representation of the system components and its control sequence coupled with a simplified thermal load. The model is then extended with fault models that resemble commonly encountered component faults to generate operation data in faulty scenarios. The proposed framework is validated in a real use case with the calibration of an air-handling unit system and its control sequence using 15 days of data gathered from the building management system. The results obtained from the injection of seven typical faults are then compared with the fault-free simulations to highlight and validate their impact on key operating variables. The results prove the capabilities of this methodology to support database generation in the fields of fault detection and advanced maintenance of buildings’ air conditioning systems for building replicas.

Reconstructing the Ocean State Using Argo Data and a Data-Driven Method

Abstract Over the past two decades, Argo profiles have provided unprecedented insight into the global patterns of space and time variability of ocean temperature and salinity, significantly reducing associated uncertainties. However, analyzing such assessments during the pre-Argo period remains challenging due to the scarcity of observations in many regions. From the Argo period, a set of dominant three-dimensional patterns can be estimated using Empirical Orthogonal Function (EOF) analysis, which helps to fill in observational gaps. From the associated principal components, temporal fluctuations can be observed, aiming to build a catalog of possible ocean state trajectories. To map pre-Argo observations, EOFs are used in a data assimilation framework that uses this catalog to feed an analog prediction and provide reanalysis. In this study, a new data-driven interpolation method called RedAnDA (Reduced-space Analog Data Assimilation) was tested in the tropical Pacific Ocean. RedAnDA was first validated through an Observing System Simulation Experiment (OSSE) approach, using synthetic observations extracted from a model simulation. It was subsequently applied to a real historical dataset and compared with other available reanalysis products. Overall, the reconstructed temperature field showed variability consistent with the OSSE true field and other reanalysis products in the real data application. Further improvements are needed to optimally estimate uncertainty, but RedAnDA already combines valuable information about state predictability, observational sampling, and unresolved scale issues.

A comparative study of the blue comets C/1908 R1 (Morehouse) and C/2016 R2 (Pan-STARRS)

The long-period comet C/1908 R1 (Morehouse) is distinguished by its early spectroscopic tail photography, which uncovered notably intense emission bands of N2+ and CO+ similar to the unusual characteristics of the atypical blue comet C/2016 R2 (Pan-STARRS). To probe potential parallels with C/2016 R2 further, we revisited the historical spectroscopic plates of C/1908 R1 while leveraging the New Astrometric Reduction of Old Observations (NAROO) project's advanced sub-micrometric scanner. We first reviewed the intensity ratio method, followed by a comprehensive spectroscopic analysis of the original historical plates to determine the comet's composition. Our analysis also encompassed an evaluation of C/1908 R1's dynamic trajectory using an N-body integrator and a detailed examination of tail morphology records. Our findings suggest that C/1908 R1 experienced no significant close encounters as it crossed the inner Solar System, anchoring its origins directly in the Oort Cloud and allowing us to ascertain that this was its inaugural voyage near the Sun. We determined a N2+ CO+ ratio of $ along with a dust-poor composition, particularities it shares with C/2016 R2. Moreover, by synthesizing observations of the tail's structure over the three-month period of visibility, we uncovered a link between tail dislocation events and aurora borealis sightings on Earth. This association underscores the comet tail’s heightened sensitivity to solar wind fluctuations due to its volatile makeup. The comet C/1908 R1 (Morehouse) emerges as one of the most unaltered relics of our Solar System's formation, offering another instance of a C/2016 R2-analogous comet. This underscores the importance of preserving and reexamining historical astronomical datasets, not only for historical significance but as a critical resource for contemporary scientific advancement.

MODIS Evapotranspiration Forecasting Using ARIMA and ANN Approach at a Water-Stressed Irrigation Scheme in South Africa

The forecasting of evapotranspiration (ET) in some water-stressed regions remains a major challenge due to the lack of reliable and sufficient historical datasets. For efficient water balance, ET remains the major component and its proper forecasting and quantifying is of the utmost importance. This study utilises the 18-year (2001 to 2018) MODIS ET obtained from a drought-affected irrigation scheme in the Eastern Cape Province of South Africa. This study conducts a teleconnection evaluation between the satellite-derived evapotranspiration (ET) time series and other related remotely sensed parameters such as the Normalised Difference Vegetation Index (NDVI), Normalised Difference Water Index (NDWI), Normalised Difference Drought Index (NDDI), and precipitation (P). This comparative analysis was performed by adopting the Mann–Kendall (MK) test, Sequential Mann–Kendall (SQ-MK) test, and Multiple Linear Regression methods. Additionally, the ET detailed time-series analysis with the Keiskamma River streamflow (SF) and monthly volumes of the Sandile Dam, which are water supply sources close to the study area, was performed using the Wavelet Analysis, Breaks for Additive Seasonal and Trend (BFAST), Theil–Sen statistic, and Correlation statistics. The MODIS-obtained ET was then forecasted using the Autoregressive Integrated Moving Average (ARIMA) and Artificial Neural Networks (ANNs) for a period of 5 years and four modelling performance evaluations such as the Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and the Pearson Correlation Coefficient (R) were used to evaluate the model performances. The results of this study proved that ET could be forecasted using these two time-series modeling tools; however, the ARIMA modelling technique achieved lesser values according to the four statistical modelling techniques employed with the RMSE for the ARIMA = 37.58, over the ANN = 44.18; the MAE for the ARIMA = 32.37, over the ANN = 35.88; the MAPE for the ARIMA = 17.26, over the ANN = 24.26; and for the R ARIMA = 0.94 with the ANN = 0.86. These results are interesting as they give hope to water managers at the irrigation scheme and equally serve as a tool to effectively manage the irrigation scheme.

Next-Generation Sequencing Methods for Near-Real-Time Molecular Epidemiology of HIV and HCV.

The World Health Organisation has set targets of reducing the transmission of new hepatitis C (HCV) infections by 90%, and ending human immunodeficiency virus-1 (HIV) as a public health threat, by 2030. To achieve this, efficient and timely viral surveillance, and effective public health interventions, are required. Traditional epidemiological methods are largely dependent on the recognition of incident cases with symptomatic illness; acute HIV and HCV infections are commonly asymptomatic, which may lead to delays in the recognition of such new infections. Instead, for these viruses, molecular epidemiology may improve the detection of, and response to, clusters of viral transmission. Molecular epidemiology using historical datasets has highlighted key populations that may have benefitted from a timely intervention. Similar analyses performed on contemporary samples are needed to underpin the 2030 targets, but this requires the generation of a cohesive dataset of viral genome sequences in near-real-time. To generate such data, methodologies harnessing next-generation sequencing (NGS) should be utilised. Here we discuss the opportunity presented by NGS for public health surveillance of HIV and HCV, and discuss three methods that can generate sequences for such analysis. These include full-length genome amplification, utilised for analysis of HCV in the research space; tiling PCR, which was the method of choice for many diagnostic laboratories in the SARS-CoV-2 pandemic; and bait-capture hybridisation, which has been utilised in local HIV outbreaks. These techniques could be applied for near-real-time HIV and HCV surveillance, informing public health strategies that will be key to achieving 2030 targets.

Metaheuristic optimization of extreme gradient boosting machine for enhanced prediction of lateral strength of reinforced concrete columns under cyclic loadings

The estimation of lateral strength in reinforced concrete (RC) columns subjected to cyclic loads is crucial in structural design. The failure of RC columns under lateral forces can lead to catastrophic structural collapses. This fact emphasizes the need for accurate assessments of their dynamic behavior. This paper proposes a data-driven model for estimating the lateral strength of RC columns. A historical dataset comprising 12 predictor variables and 247 samples has been compiled to train and validate of the proposed approach. The extreme gradient boosting machine (XGBoost) is employed to establish a predictive relationship between the lateral strength of RC columns and their influencing factors. Since model selection is critical for constructing a reliable prediction mode, this study relies on utilizing metaheuristic approaches, including Genetic Algorithm, Particle Swarm Optimization, Artificial Bee Colony, and Ant Colony Optimization, to optimize the performance of the XGBoost model. Experimental results show that the integration of Ant Colony Optimization and XGBoost can help attain outstanding prediction accuracy with a correlation of determination (R2) of 0.95. Additionally, an asymmetric squared error loss function is utilized to reduce overestimations by 12 %. The newly developed method can be utilized in practical applications where reliable predictions of the lateral strength of RC columns under cyclic loads are required.

A Multi-Decadal Hourly Coincident Wind and Solar Power Production Dataset for the Contiguous United States

As renewable energy continues its rapid expansion in the Unites States, multi-decadal hourly datasets of electricity production are needed to asses reliability and resource adequacy of power grids. Recent years have seen the release of grid-cell-level simulated meteorological variables, however these are not extended to the power domain, are not developed from a dynamically consistent numerical weather model, and only cover a historical baseline of less than a decade. To fill this gap, this work provides a dataset of 43 years of coincident plant-level wind and solar power production data. The dataset is designed to be aggregated to appropriate scales of interest for bulk system studies such as Balancing Authorities (BAs), states, and nodes of a production cost model. The dataset covers every plant in the contiguous U.S. that is reported in the U.S. Energy Information Administration (EIA) Form 860 as of 2020. When compared with the EIA-923 monthly generation, we find minimal bias (less than 5%). When compared with BA-reported hourly generation, we find low bias in solar (less than 7%), and slight underdispersion in wind. This coincident multi-decadal historical dataset provides a documented and evaluated multi-resource baseline for studies on reliability, resource adequacy, climate change impacts, and characterization of emergent climate threats on renewable resources.

A “Region-Specific Model Adaptation (RSMA)”-Based Training Data Method for Large-Scale Land Cover Mapping

An accurate and historical land cover monitoring dataset for Alaska could provide fundamental information for a range of studies, such as conservation habitats, biogeochemical cycles, and climate systems, in this distinctive region. This research addresses challenges associated with the extraction of training data for timely and accurate land cover classifications in Alaska over longer time periods (e.g., greater than 10 years). Specifically, we designed the “Region-Specific Model Adaptation (RSMA)” method for training data. The method integrates land cover information from the National Land Cover Database (NLCD), LANDFIRE’s Existing Vegetation Type (EVT), and the National Wetlands Inventory (NWI) and machine learning techniques to generate robust training samples based on the Anderson Level II classification legend. The assumption of the method is that spectral signatures vary across regions because of diverse land surface compositions; however, despite these variations, there are consistent, collective land cover characteristics that span the entire region. Building upon this assumption, this research utilized the classification power of deep learning algorithms and the generalization ability of RSMA to construct a model for the RSMA method. Additionally, we interpreted existing vegetation plot information for land cover labels as validation data to reduce inconsistency in the human interpretation. Our validation results indicate that the RSMA method improved the quality of the training data derived solely from the NLCD by approximately 30% for the overall accuracy. The validation assessment also demonstrates that the RSMA method can generate reliable training data on large scales in regions that lack sufficient reliable data.

Quantifying Climate Change Variability for the Better Management of Water Resources: The Case of Kobo Valley, Danakil Basin, Ethiopia

Alterations in the hydrological cycle due to climate change are one of the key threats to the future accessibility of natural resources. This study used 12 GCM climate models from CMIP6 to evaluate future climate change scenarios by applying model performance measures and trend analysis in Kobo Valley, Ethiopia. The models were ranked based on their ability to analyze the historical datasets. The result of this study showed that the outputs of the FIO-ESM-2-0 CIMP6 model had a good overall ranking for both precipitation and temperature. After bias correction of the model-based projections with the observed data, the average annual precipitation in the average scenario (SSP2-4.5) decreased by 4.4% and 13% in 2054 and 2084, respectively. Similarly, in the worst-case scenario (SSP5-8.5), by the end of 2054 and 2084, decreases of 4% and 12.8%, respectively, were predicted. The average annual maximum temperature under the SSP2-4.5 scenario increased by 1.5 °C in 2054 and by 2.1 °C in 2084. The average annual maximum temperature under the worst-case (SSP5-8.5) scenario increased by 1.7 °C in 2054 and by 3.2 °C in 2084. In the middle scenario (SSP4.5), the average annual minimum temperature increased by 2.2 °C in 2054 and by 3 °C in 2084. The average annual minimum temperature under the worst-case (SSP5-8.5) scenario increased by 2.6 °C in 2054 and by 4.3 °C in 2084. The seasonal variability in precipitation in the studied valley will decrease in the winter and increase in the summer. A decrease in precipitation combined with an increase in temperature will strengthen the risk of drought events in the future.

Examining the effect of soil organic carbon on major Canadian Prairie crop yields with predictive soil mapping

AbstractMaintaining soil organic carbon (SOC) is critical for global food security as it is essential for soil functions that sustain crop yields. There has been an increase in predictive soil mapping, which when combined with extensive crop yield datasets, enables a better understanding of crop yield and SOC relationships. This study focused on updating maps of SOC content in Saskatchewan using recently digitized historical SOC datasets and predictive soil mapping, and using the maps to examine the relationship between SOC and crop yield. A database of 5014 SOC values was used to map SOC contents using a Random Forest model and a range of environmental covariates. The final SOC model had a R2 of 0.48, root mean square error of 0.98%, concordance correlation coefficient of 0.67, and a bias of 0.12%. The relationship between mapped SOC values and crop yield data, with 100,000–200,000 records depending on crop type, was then assessed using a linear mixed effects model after normalizing the data by rural municipality to remove broad‐scale climate effects. Overall, an increase in SOC by 1% led to an increase on average of 263 kg ha−1 for wheat (Triticum aestivum L.), 293 kg ha−1 for barley (Hordeum vulgare L.), 133 kg ha−1 for canola (Brassica napus L.), and 135 kg ha−1 for field peas (Pisum sativum L.). These results show that increasing SOC was associated with greater yields for four major crops in Saskatchewan, with the largest gains occurring when the initial SOC contents are lower.

FROM CALENDAR TIME TO BUSINESS TIME: THE CASE OF COMMODITY MARKETS

We address the problem of modeling commodity forward curves, while preserving empirical features observed in commodity markets. By letting a commodity market to “ live” in the tempo of a business rather than calendar clock, we create a model with a rich but realistic set of features, such as stochastic volatility, stochastic rate of mean reversion and various shapes of forward curves such as backwardation and contango. The model, when applied to extensive historical datasets of crude oil and natural gas forward curves, shows a remarkably good fit to the observed futures prices, also in periods of high volatility and negative prices. The model is developed in such a way that it can be used for a wide variety of applications, ranging from exotic derivatives pricing to risk management of commodity portfolios.

CGAformer: Multi-scale feature Transformer with MLP architecture for short-term photovoltaic power forecasting

Accurately predicting the output power of photovoltaic (PV) systems is an effective means to ensure the reliable and economical operation of grid-connected PV systems. Aiming at the characteristics of PV power generation such as strong volatility, high intermittency and obvious periodicity, a hybrid model named CGAformer based on One-Dimensional Convolutional Neural Networks (CNN1D), Global Additive Attention (GADAttention), and Auto-Correlation is proposed for short-term PV power generation prediction. The model uses CNN1D to extract local features and obtains global weights by improving the GADAttetion obtained by additive attention. Auto-Correlation integrates local features and global weights and identifies repeated patterns in the sequence to obtain highly coupled multi-scale features, and finally generates the final prediction results through Multilayer Perceptron (MLP). In order to verify the effectiveness of the model, this paper uses a historical dataset from a PV system located in Uluru, Australia for sufficient experiments. In the comparative experiments, The overall average RMSE and MAE of CGAformer are improved by 6.82% and 20.46% respectively compared with long short-term memory (LSTM). In addition, ablation experiments and seasonal analysis are used to verify the effectiveness of the model and its excellent generalization ability for different seasons.

NSEMBLE FINE TUNED MULTI LAYER PERCEPTRON FOR PREDICTIVE ANALYSIS OF WEATHER PATTERNS AND RAINFALL FORECASTING FROM SATELLITE DATA

The accurate prediction of weather patterns and rainfall forecasting is critical for various sectors, including agriculture, disaster management, and water resource planning. Traditional models often struggle to capture the complex interactions between atmospheric variables, particularly when integrating diverse types of satellite data (binary, categorical, and numerical). To address this challenge, an ensemble fine-tuned multi-layer perceptron (MLP) model is developed, combining the strengths of multiple machine learning techniques for more robust predictions. The primary problem is the difficulty in handling mixed data types while maintaining high prediction accuracy. Satellite data, including binary indicators (e.g., cloud presence), categorical features (e.g., cloud types), and numerical variables (e.g., temperature, humidity, and wind speed), provide rich information but require specialized processing for effective forecasting. The proposed method involves fine-tuning an ensemble of MLP models with backpropagation, dropout regularization, and batch normalization to reduce overfitting and enhance generalization. The ensemble integrates predictions from individual MLP models, each trained on different subsets of features (binary, categorical, numerical). This technique allows the model to leverage complementary strengths and produce more accurate rainfall forecasts. Satellite data is preprocessed and normalized before training, and categorical variables are one-hot encoded to ensure compatibility with the MLP architecture. Results from testing on historical satellite weather datasets demonstrate significant improvements in forecast accuracy. The ensemble MLP achieved an accuracy of 91.3%, with a precision of 90.7%, recall of 89.5%, and an F1-score of 90.1%. The model performed exceptionally well in identifying critical rainfall events, reducing false positives by 12% compared to traditional models.

A critical reflection on attempts to machine-learn materials synthesis insights from text-mined literature recipes.

Synthesis of predicted materials is the key and final step needed to realize a vision of computationally accelerated materials discovery. Because so many materials have been previously synthesized, one would anticipate that text-mining synthesis recipes from the literature would yield a valuable dataset to train machine-learning models that can predict synthesis recipes for new materials. Between 2016 and 2019, the corresponding author (Wenhao Sun) participated in efforts to text-mine 31 782 solid-state synthesis recipes and 35 675 solution-based synthesis recipes from the literature. Here, we characterize these datasets and show that they do not satisfy the "4 Vs" of data-science-that is: volume, variety, veracity and velocity. For this reason, we believe that machine-learned regression or classification models built from these datasets will have limited utility in guiding the predictive synthesis of novel materials. On the other hand, these large datasets provided an opportunity to identify anomalous synthesis recipes-which in fact did inspire new hypotheses on how materials form, which we later validated by experiment. Our case study here urges a re-evaluation on how to extract the most value from large historical materials-science datasets.