Linear Regression Trees Research Articles

A decision tree model for accurate prediction of sand erosion in elbow geometry

Erosion of piping components, e.g., elbows, is a hazardous phenomenon that frequently occurs due to sand flow with fluids during petroleum production. Early prediction of the sand's erosion rate (ER) is essential for ensuring a safe flow process and material integrity. Some models have been applied to determine the ER of the sand in the literature. However, these models have been created based on specific data to require a model for application to wide-range data. Moreover, the previous models have not studied relationships between independent and dependent variables. Thus, this research aims to use machine learning techniques, namely linear regression and decision tree (DT), to predict the ER robustly. The optimum model, the DT model, was evaluated using various trend analysis and statistical error analyses (SEA) techniques, namely the correlation coefficient (R). The evaluation results proved proper physical behavior for all independent variables, along with high accuracy and the DT model robustness. The proposed DT method can accurately predict the ER with R of 0.9975, 0.9911, 0.9761, and 0.9908, AAPRE of 5.0%, 6.27%, 6.26%, and 5.5%, RMSE of 2.492E-05, 6.189E-05, 9.310E-05, and 5.339E-05, and STD of 13.44, 6.66, 8.01, and 11.44 for the training, validation, testing, and whole datasets, respectively. Hence, this study delivers an effective, robust, accurate, and fast prediction tool for ER determination, significantly saving the petroleum industry's cost and time.

Classification and quantification of sucrose from sugar beet and sugarcane using optical spectroscopy and chemometrics.

Sucrose, obtained from either sugar beet or sugarcane, is one of the main ingredients used in the food industry. Due to the same molecular structure, chemical methods cannot distinguish sucrose from both sources. More practical and affordable methods would be valuable. Sucrose samples (cane and beet) were collected from nine countries, 25% (w/w) aqueous solutions were prepared and their absorbances recorded from 200 to 1380nm. Spectral differences were observable in the ultraviolet-visible (UV-Vis) region from 200 to 600nm due to impurities in sugar. Linear discriminant analysis (LDA), classification and regression trees, and soft independent modeling of class analogy were tested for the UV-Vis region. All methods showed high performance accuracies. LDA, after selection of five wavelengths, gave 100% correct classification with a simple interpretation. In addition, binary mixtures of the sugar samples were prepared for quantitative analysis by means of partial least squares regression and multiple linear regression (MLR). MLR with first derivative Savitzky-Golay were most acceptable with root mean square error of cross-validation, prediction, and the ratio of (standard error of) prediction to (standard) deviation values of 3.92%, 3.28%, and 9.46, respectively. Using UV-Vis spectra and chemometrics, the results show promise to distinguish between the two different sources of sucrose. An affordable and quick analysis method to differentiate between sugars, produced from either sugar beet or sugarcane, is suggested. This method does not involve complex chemical analysis or high-level experts and can be used in research or by industry to detect the source of the sugar which is important for some countries' agricultural policies.

Predicting fish species richness and abundance in the Lower Mekong Basin

Predictive models are widely used to investigate relationships between the distribution of fish diversity, abundance, and the environmental conditions in which they inhabit, and can guide management actions and conservation policies. Generally, the framework to model such relationships is established; however, which models perform best in predicting fish diversity and abundance remain unexplored in the Mekong River Basin. Here, we evaluated the performance of six single statistical models namely Generalized Linear Model, Classification and Regression Tree, Artificial Neural Network, k-Nearest Neighbor, Support Vector Machine and Random Forest in predicting fish species richness and abundance in the Lower Mekong Basin. We also identified key variables explaining variability and assessed the variable’s sensitivity in prediction of richness and abundance. Moreover, we explored the usefulness of an ensemble modeling approach and investigated if this approach improved model performance. Our results indicated that, overall, the six single statistical models successfully predicted the fish species richness and abundance using 14 geo-hydrological, physicochemical and climatic variables. The Random Forest model consistently out-performed all single statistical models for predicting richness (R2 = 0.85) and abundance (R2 = 0.77); whereas, Generalized Linear Model performed the worst of all models (R2 = 0.60 and 0.56 for richness and abundance). The most important predictors of variation in both richness and abundance included water level, distance from the sea and alkalinity. Additionally, dissolved oxygen, water temperature and total nitrate were important predictors of species richness, while conductivity was important for fish abundance. We found that species richness increased with increasing water level, dissolved oxygen and water temperature, but decreased with increasing distance from the sea, alkalinity and total nitrate. Fish abundance increased with conductivity, but decreased with increasing distance from the sea, water level and alkalinity. Finally, our results highlighted the usefulness of ensemble modeling (R2 = 0.90 and 0.85 for richness and abundance) for providing better predictive power than any of the six single statistical models. Our results can be used to support Mekong River management, particularly fisheries in the context of contemporary regional and global changes.

A Comparative Study of HARR Feature Extraction and Machine Learning Algorithms for Covid-19 X-Ray Image Classification

In this study, we investigated how effectively COVID-19 image categorization using Harr feature extraction and machine learning algorithms. We were particularly interested in the effectiveness of these algorithms. A dataset of 500 X-ray scans, equally split between 250 COVID-19-positive cases and 250 healthy controls, served as the basis for our study. K-nearest neighbors,decision tree, Linear regression, support vector machine, regression, classification, naive Bayes,random forest, as well as linear discriminant analysis were among the seven machine-learning approaches used to categorize the photos. With the use of Harr feature extraction, the features of the pictures were extracted. We studied the efficacy of COVID-19 X-ray images for classification utilizing the combination of machine learning as well as the Harr feature extraction methods in the present investigation due to their effectiveness. We searched a database of 500 X-rays for this investigation, dividing them equally between groups of 250 patients with COVID-19-positive cases and 250 healthy people. Following that, the images were examined using seven various machine learning approaches for recognition. These methods included naive Bayes, linear discriminant analysis, random forests, classification,k-nearest neighbors, and regression trees. The information from the photos was gathered using the Harr feature extraction method. The effectiveness of the algorithms was evaluated with the help of a variety of metrics, such asF1 score, precision,accuracy, recall, the area under the ROC curve, and the region of interest curve. According to our research, the Support Vector Machine algorithm had the highest accuracy, at 77%, while the Naive Bayes approach had the lowest accuracy, at 58%. By using machine learning and Harr feature extraction approaches, the Random Forest method yields the best results, based on our research. The development of future COVID-19 X-ray image-based automated diagnostic systems may be influenced by these findings. Results from the suggested model were comparable to those of cutting-edge models trained using transfer learning techniques. The proposed model's main advantage is that it has ten times fewer parameters than the most advanced models.A receiver operating characteristic (ROC) curve's F1 score, and the algorithms' accuracy, precision, the area under the curve, and recall were all used as metrics. According to our findings, the Naive Bayes method gained the least accuracy (58%) and the Support Vector Machine method produced the highest accuracy (77%) when used. Our results reveal that employing Harr feature extraction and machine learning techniques, the Random Forest strategy is the most successful way to recognize COVID-19 X-ray pictures. These findings may be pertinent to the development of automated COVID-19 diagnosis tools relying on X-ray images. The recommended model produced results that were competitive when measured against cutting-edge models trained using transfer learning techniques. The suggested model employs 10 times fewer parameters than the most advanced models, which is its key selling point.

Model Sistem Penunjang Keputusan Untuk Pengelolaan Pembiayaan Nasabah Di BPR Sehat Sejahtera Universitas Pamulang

Abstract
 
 One of the BPRs that is currently developing in South Tangerang is BPR Sehat Sejahtera. BPR Sehat Sejahtera is part of the Pamulang University Sasmita Group, whose function as a bank is to channel financing to community businesses so that they are able to support the country's economy. To distribute financing, BPR managers are expected to be able to conduct financing analysis to reduce the risk of financing failure. Until now, financing analysis carried out by BPRs tends to be manual and requires quite a long time to make financing decisions. One method that can be used to assist in financing analysis is the credit scoring model. To facilitate decision making, an SPK was formed with a credit scoring model to provide several choices regarding customer financing applications. The DSS development process follows the traditional software development stages, namely: (1) planning, (2) analysis, (3) design and (4) implementation. SPK is developed using a spreadsheet model with a dialog display that supports user needs. The credit scoring model developed uses several methods, namely discriminant analysis, multiple linear regression, logistic regression, and decision trees. Each method will produce a different credit scoring model to be included in the SPK model base. The SPK prototype developed resulted in several financing analyzes according to the credit scoring model used. The prediction accuracy shown by credit scoring models ranges from 70% -80%. This method is capable of producing a model that provides a fairly high predictive accuracy. Meanwhile, the decision tree method is still not optimal in producing a credit scoring model.
 
 Keywords: financing analysis, discriminant analysis, BPR, credit scoring, decision tree, multiple linear regression, logistic regression, decision support system (DSS)

Identifying potential significant factors impacting zero-inflated proportion data.

Classical supervised methods like linear regression and decision trees are not completely adapted for identifying impacting factors on a response variable corresponding to zero-inflated proportion data (ZIPD) that are dependent, continuous and bounded. In this article we propose a within-block permutation-based methodology to identify factors (discrete or continuous) that are significantly correlated with ZIPD, we propose a performance indicator quantifying the percentage of correlation explained by the subset of significant factors, and we show how to predict the ranks of the response variables conditionally on the observation of these factors. The methodology is illustrated on simulated data and on two real data sets dealing with epidemiology. In the first data set, ZIPD correspond to probabilities of transmission of Influenza between horses. In the second data set, ZIPD correspond to probabilities that geographic entities (eg, states and countries) have the same COVID-19 mortality dynamics.

Automated neural network forecast of PM2.5 concentration

Abstract Particulate Matter 2.5 (PM2.5) is a major contributor to air pollution and its exposure has substantial health consequences. As a result, precise prediction of PM2.5 concentrations is required in order to establish emission reduction strategies for air quality management. The article presents an Artificial Neural Network (ANN) model to forecast PM2.5 levels in a particular region. The model uses data such as air temperature, carbon monoxide, nitric oxide, nitrogen dioxide, ozone, suspended particles, rainfall, relative humidity, sulfur dioxide, wind direction and wind speed to predict PM2.5 concentrations in the air accurately. The model’s efficacy is evaluated using statistical measures such as the Coefficient of Determination, the Root Mean Squared Error and the Mean Absolute Error. The study results indicate that the ANN model outperforms more traditional statistical models, with R2 values of 0.987, which is higher than the values achieved by the Linear Regression and Decision Tree Regressor models, which are 0.88 and 0.89 respectively. The study’s findings have significant implications for public health and environmental policy, as they can provide more accurate and rapid statistics on air quality. The ability to forecast PM2.5 concentrations can help policymakers and health professionals take proactive measures to mitigate the impact of air pollution on public health.

BMSP-ML: big mart sales prediction using different machine learning techniques

<span lang="EN-US">Variations in sales over time is the main issue faced by many retailers. To overcome this problem, we attempt to predict the sales by comparing the previous sales data of different stores. Firstly, the primary task is to recognize the pattern of the factors that help to predict sales. This study helps us understand the data and predict sales using many machines learning models. This process gets the data and beautifies the data by imputing the missing values and feature engineering. While solving this problem, predicting the monthly sales value is significant in the study. In addition, an essential element is to clear the missing data and perform proper feature engineering to better understand them before applying them. The experimental results show that the random forest predictor has outperformed ridge regression, linear regression, and decision tree models among the four machine learning techniques implemented in this study. The performance of the proposed models has been evaluated using root mean square error (RMSE).</span>

Performance Prediction Model Development for Solar Box Cooker Using Computational and Machine Learning Techniques

Abstract The development of prediction models for solar thermal systems has been a research interest for many years. The present study focuses on developing a prediction model for solar box cookers (SBCs) through computational and machine learning (ML) approaches. The prime objective is to forecast cooking load temperatures of SBC through ML techniques such as random forest (RF), k-nearest neighbor (k-NN), linear regression (LR), and decision tree (DT). ML is a commonly used form of artificial intelligence, and it continues to be popular and attractive as it finds new applications every day. A numerical model based on thermal balance is used to generate the dataset for the ML algorithm considering different locations across the world. Experiments on the SBC in Indian weather conditions are conducted from January through March 2022 to validate the numerical model. The temperatures for different components obtained through numerical modeling agree with experimental values with less than 7% maximum error. Although all the developed models can predict the temperature of cooking load, the RF model outperformed the other models. The root-mean-square error (RMSE), determination coefficient (R2), mean absolute error (MAE), and mean square error (MSE) for the RF model are 2.14 (°C), 0.992, 1.45 (°C), and 4.58 (°C), respectively. The regression coefficients indicate that the RF model can accurately predict the thermal parameters of SBCs with great precision. This study will inspire researchers to explore the possibilities of ML prediction models for solar thermal conversion applications.

Machine learning to improve the performance of anomaly-based network intrusion detection in big data

With the rapid growth of digital technology communications are overwhelmed by network data traffic. The demand for the internet is growing every day in today's cyber world, raising concerns about network security. Big Data are a term that describes a vast volume of complicated data that is critical for evaluating network patterns and determining what has occurred in the network. Therefore, detecting attacks in a large network is challenging. Intrusion detection system (IDS) is a promising cybersecurity research field. In this paper, we proposed an efficient classification scheme for IDS, which is divided into two procedures, on the CSE-CIC-IDS-2018 dataset, data pre-processing techniques including under-sampling, feature selection, and classifier algorithms were used to assess and decide the best performing model to classify invaders. We have implemented and compared seven classifier machine learning algorithms with various criteria. This work explored the application of the random forest (RF) for feature selection in conjunction with machine learning (ML) techniques including linear regression (LR), k-Nearest Neighbor (k-NN), classification and regression trees (CART), Bayes, RF, multi layer perceptron (MLP), and XGBoost in order to implement IDSS. The experimental results show that the MLP algorithm in the most successful with best performance with evaluation matrix.

Using a variety of machine learning approaches to predict and map topsoil pH of arable land on a regional scale

AbstractIn order to accurately predict soil properties, various machine learning (ML) approaches and hybrid models constructed by integrating ML into regression kriging framework were used to predict and map arable land topsoil pH in Henan province, central China. Random forest (RF), cubist (Cu), support vector machine, artificial neural network, multiple linear regression, classification and regression trees (CART) and their hybrid models were compared for pH accuracy prediction. Among all standalone ML models, RF had the best predictive performance, in terms of the metrics employed in this study, followed by Cu, and CART was the worst. Compared with their ML counterparts, hybrid models could improve the accuracy of topsoil pH prediction to various extents. The accuracy improvement of the hybrid models constructed based on the simple ML was much greater than that based on the complex ensemble ML. Except for artificial neural network kriging , there was no significant difference between different hybrid models in the predicted results of topsoil pH. The outputs from the best predictive models showed that weak acidic soils and weak alkaline soils were the dominant arable soils in the study region, accounting for more than 30% and more than 50% of the total arable land area respectively, the topsoil pH of arable land in the north of the study area is generally higher than that in the south. Boruta variable selection revealed that altitude, climatic covariates closely related to soil moisture availability and some soil properties were the most critical factors affecting and controlling the topsoil pH of arable land.

On cobalt zinc boro sodium fluoride glasses doped with Y2O3: Synthesis, artificial intelligence density prediction and dielectric spectroscopy

Artificial intelligence density prediction and dielectric spectroscopy of five glass samples with chemical compositions 64.9B2O3-25NaF-10ZnO-0.1CoO-xY2O3: x = 0.0 (Y0.0), 0.2 (Y0.2), 0.4 (Y0.4), 0.8 (Y0.8), and 1.0 (Y1.0) mol% have been investigated. Fabrication process has been achieved by melting quenching method at melting temperature 1050–1100 °C. Density of the Y0.0-Y1.0 samples experimentally was measured and varied from 2.945 to 3.016 g/cm3. The chemical composition of the prepared glasses successfully mapped with various machine learning (ML) algorithms to predict densities using 10,000 glass data set. The density prediction done by various machine learning (ML) techniques like the logistic regression, linear regression, and decision tree (DT), random forest regression (RFR), and artificial neural network (ANN). A decision tree algorithm being fitted to the training set, estimating the test outcome utilizing the confusion matrix. The accuracy of 78.261 as well as a classification error equal 0.392 obtained for our data set which was best values to predict density for various ML techniques. The logistic regression also predicting successfully the glass data with R2 0.850 compare to linear regression. The random forest regression dominating in density prediction in current study with maximum R2 values 0.987. In addition, the dielectric spectroscopy of the prepared Y0.0-Y1.0 glasses is examined through various frequencies at room temperature (RT) to investigate the structure changes by yttrium doping on dielectric spectroscopy items. Results revealed that the insertion of yttrium ion causes an enhancement in dielectric constant and the dependency of dielectric loss (tan δ) on applied frequency has relaxation peaks at around 700 and 7000 Hz.

Interpretable machine learning models for the estimation of seismic drifts in CLT buildings

An accurate estimation of drift demands is crucial for designing and assessing structures under seismic loads. Given the novelty of massive timber buildings, predictive models for the estimation of drifts in mid-to high-rise CLT structures are lacking, particularly in the form of simple models suitable for preliminary design evaluations or regional seismic assessments. In this paper, we present and compare several Machine Learning (ML) models for the estimation of peak inter-storey and roof drifts in multi-storey Cross-Laminated Timber (CLT) walled structures. The ML techniques used include: Multiple Linear Regression, Regression Trees, Random Forest, K-nearest Neighbor, and Support Vector Regression. To this end, 69 structures spanning mid-rise to tall timber buildings are subjected to a large collection of acceleration records and used to create the training and testing datasets. Different structural configurations and behaviour factors, related to the assumed energy dissipation capacity of the buildings, are considered. A diversity of feature selection techniques informs our choice of parameters to the reduced input space leading to a set of six most efficient features: the spectral acceleration at the building's fundamental period (Sa(T1)), the Peak Ground Velocity (PGV), tuning ratio (T1/Tm), behaviour factor (q), wall height (Hw), and the wall subdivision ratio (Wr). After verifying the high accuracy of our model predictions, the SHapley Additive exPlanation method (SHAP) is used to gain insight into the influence of key input features on the ML model outputs. Finally, our ML drift estimations are compared against previous proposals and design code assumptions, and the potential causes of disagreement are discussed.

Digital assessments of soil organic carbon storage using digital maps provided by static and dynamic environmental covariates

AbstractUnderstanding the vertical and lateral distribution of soil organic carbon (SOC) and soil organic carbon density (SOCD) is indispensable for soil use and environmental management because of their vital role in soil quality assessments. Primarily, they are needed in calculating soil organic carbon storage (SOCS). The objective of this research was to provide digital maps of SOC and SOCD variation as well as their uncertainties at multiple standardized depths (H1: 0–5, H2: 5–15, H3: 15–30, H4: 30–60 and H5: 60–100 cm) using a parsimonious model with optimized terrain‐related attributes and satellite‐derived data. SOCS were evaluated at soil subgroup levels. An area of about 808 km2 with varying elevation, plant cover and lithology from the Miandoab region, West Azerbaijan Province, Iran was selected as a case study area. A total of 386 soil samples were collected from 104 profiles comprising various soil genetic horizons. A continuous spline function was then fitted to the target properties in advance of creating a dataset at five standard depth intervals (following the GlobalSoilMap project). These were then grouped into three classes including top (H1), middle (H2, H3 and H4) and bottom (H5) depths to ease interpretation. Static and dynamic covariates (30‐m resolution) were derived from a digital elevation model (DEM) and a suite of Landsat‐8 spectral imageries, respectively. Four candidate models including stepwise multiple linear regression (SMLR), random forest (RF), cubist (CU) and extreme gradient boosting (XGBoost) Tree were tested in this study. Finally, the digital maps at 30‐m resolution of SOC and SOCD and their uncertainties were prepared using the best‐fit model and the bootstrapping method, respectively. Four soil subgroups (Gypsic Haploxerepts, Typic Calcixerepts, Typic Haploxerepts and Xeric Haplocalcids) were identified across the study area. The covariates had variable contributions on the evaluated models. The XGBoost Tree model generally outperformed other models for prediction of SOC and SOCD (R2 = 0.60, on average). Regardless of soil subgroups, the uncertainty analysis showed that the SOCD map had a low prediction interval range value indicating high accuracy. Additionally, the highest SOCS and SOCD was observed at the top followed by middle and bottom depths in the study area. All subgroups exhibited a decreasing trend of SOCD with increasing depth. A similar trend was also observed for SOCS. The highest SOCD (on average) was observed in Gypsic Haploxerepts (4.71 kg C/m2) followed by Typic Calcixerepts (4.46 kg C/m2), Typic Haploxerepts (4.45 kg C/m2) and Xeric Haplocalcids (4.40 kg C/m2). Overall, the SOCS normalized by area within soil order boundaries was greater in Inceptisols than Aridisols across the study area. The findings of this study provide critical information for sustainable management of soil resources in the area for agricultural production and environmental health in the Miandoab region of Iran.

AIR QUALITY PREDICTION USING MACHINE LEARNING ALGORITHMS

Days gone by air pollution is rapidly increasing, pollution occurs due to human activities, industrialization and burning of fossil fuels. The air is polluted by dangerous gases present in the atmosphere; the pollutants are carbon dioxide (CO2), carbon monoxide (CO), Sulphur dioxide (SO2), nitrogen dioxide (NO2), etc. The pollutants of which Particulate Matter (PM 2.5) consists of suspended particles with a diameter of less than 2.5 micrometers are considered harmful. Air pollution is a major issue that impacts humans and causes health problems including heart disease, lung cancer, and respiratory diseases such as emphysema and asthma. previous techniques such as probability, statistics, traditional methods, etc. These methods are too complex to predict. Machine learning (ML) is the best way to predict air quality. In this paper, we need to implement models that record information about air pollutant concentrations. We compare machine learning algorithms like linear regression (LR), random forest (RF), k-nearest neighbor (KNN), and decision tree regression (DT) to predict the air quality index (AQI). Key Words: Machine learning- Linear Regression, Random Forest, k-nearest neighbor, and Decision Tree regression

Prediction and Classification of User Activities Using Machine Learning Models from Location-Based Social Network Data

The current research has aimed to investigate and develop machine-learning approaches by using the data in the dataset to be applied to classify location-based social network data and predict user activities based on the nature of various locations (such as entertainment). The analysis of user activities and behavior from location-based social network data is often based on venue types, which require the input of data into various categories. This has previously been done through a tedious and time-consuming manual method. Therefore, we proposed a novel approach of using machine-learning models to extract these venue categories. In this study, we used a Weibo dataset as the main source of research and analyzed machine-learning methods for more efficient implementation. We proposed four models based on well-known machine-learning techniques, including the generalized linear model, logistic regression, deep learning, and gradient-boosted trees. We designed, tested, and evaluated these models. We then used various assessment metrics, such as the Receiver Operating Characteristic or Area Under the Curve, Accuracy, Recall, Precision, F-score, and Sensitivity, to show how well these methods performed. We discovered that the proposed machine-learning models are capable of accurately classifying the data, with deep learning outperforming the other models with 99% accuracy, followed by gradient-boosted tree with 98% and 93%, generalized linear model with 90% and 85%, and logistic regression with 86% and 91%, for multiclass distributions and single class predictions, respectively. We classified the data using our machine-learning models into the 10 classes we used in our previous study and predicted tourist destinations among the data to demonstrate the effectiveness of using machine learning for location-based social network data analysis, which is vital for the development of smart city environments in the current technological era.

Price Prediction for Starbucks Corporation Based on Random Forest, Linear Regression, and Decision Tree

The outbreak of the virus COVID-19 in 2019 has dealt an unprecedented blow to the global economy. Contemporarily, the global economy is in the recovery stage after the end of the epidemic, and the economic recession is likely to be the future trend in terms of current status. In this paper, three state-of-art multivariate supervised learning regression scenarios (i.e., Random Forest, Linear Regression, and Decision Tree) are implemented to predict the stock price of Starbucks. Through comparison, it is concluded that Linear Regression and Random Forest are the most accurate models that are expected to be used in the future. Moreover, investing in Starbucks stock is not a wise choice at this time. It is hoped to reduce investor losses by choosing the optimal model for the most accurate stock predictions and minimizing losses in a recession. These results shed light on guiding further exploration of targeted analysis of the new economic environment.

Reproducible Model Selection Using Bagged Posteriors.

Bayesian model selection is premised on the assumption that the data are generated from one of the postulated models. However, in many applications, all of these models are incorrect (that is, there is misspecification). When the models are misspecified, two or more models can provide a nearly equally good fit to the data, in which case Bayesian model selection can be highly unstable, potentially leading to self-contradictory findings. To remedy this instability, we propose to use bagging on the posterior distribution ("BayesBag") - that is, to average the posterior model probabilities over many bootstrapped datasets. We provide theoretical results characterizing the asymptotic behavior of the posterior and the bagged posterior in the (misspecified) model selection setting. We empirically assess the BayesBag approach on synthetic and real-world data in (i) feature selection for linear regression and (ii) phylogenetic tree reconstruction. Our theory and experiments show that, when all models are misspecified, BayesBag (a) provides greater reproducibility and (b) places posterior mass on optimal models more reliably, compared to the usual Bayesian posterior; on the other hand, under correct specification, BayesBag is slightly more conservative than the usual posterior, in the sense that BayesBag posterior probabilities tend to be slightly farther from the extremes of zero and one. Overall, our results demonstrate that BayesBag provides an easy-to-use and widely applicable approach that improves upon Bayesian model selection by making it more stable and reproducible.

Evaluation of pedotransfer functions to estimate some of soil hydraulic characteristics in North Africa: A case study from Morocco

Soil hydraulic properties are an important factor to optimize and adapt water management for a given crop. Pedotransfer functions (PTFs) present a solution to predict soil variables such as hydraulic properties, using fundamental soil properties. In this research, we compared two sources of soil information: iSDAsoil data and field data, in four regions in Morocco. We then used this data to evaluate published data and developed new PTFs using soil information to estimate soil gravimetric moisture content at saturation (w0), field capacity (w330) and permanent wilting point (w15000). A total of 331 samples were collected from four regions: Doukkala, Gharb-Loukous, Moulouya and Tadla. The data was divided into calibration and validation datasets. For development of different PTFs, we used simple linear regression, multiple linear regression, regression tree, Cubist algorithm, and random forest approaches. PTFs developed by Dijkerman (Geoderma, 1988, 42, 29–49) presented the best performance, showing lower RMSE, Bias and MAE compared to other PTFs. Using multiple linear regression to develop PTFs, models based on clay, silt and soil organic matter as input variables showed the best performance after calibration (R2 of 0.590, 0.785, 0.786 for w0, w330, and w15000, respectively). Regarding the techniques based on machine learning, random forest showed the best performance after calibration compared with other algorithms (R2 of 0.930, 0.955, 0.954 for w0, w330, and w15000, respectively). PTFs represent a low cost and easy technique to estimate soil hydraulic properties, to improve water management efficiency for the farmers.

Analysis of Enrollment Criteria in Secondary Schools Using Machine Learning and Data Mining Approach

Out-of-school children (OSC) surveys are conducted annually throughout Pakistan, and the results show that the literacy rate is increasing gradually, but not at the desired speed. Enrollment campaigns and targets system of enrollment given to the schools required a valuable model to analyze the enrollment criteria better. In existing studies, the research community mainly focused on performance evaluation, dropout ratio, and results, rather than student enrollment. There is a great need to develop a model for analyzing student enrollment in schools. In this proposed work, five years of enrollment data from 100 schools in the province of Punjab (Pakistan) have been taken. The significant features have been extracted from data and analyzed through machine learning algorithms (Multiple Linear Regression, Random Forest, and Decision Tree). These algorithms contribute to the future prediction of school enrollment and classify the school’s target level. Based on these results, a brief analysis of future registrations and target levels has been carried out. Furthermore, the proposed model also facilitates determining the solution of fewer enrollments in school and improving the literacy rate.