Multi-linear Regression Algorithm Research Articles

Hardness prediction in the upsetting process of Al%ZrO2—an approach to machine learning using regression and classification models

The current study focuses on the prediction of metal hardness distribution in upsetting tests for different compositions of ZrO2 embedded with an aluminum matrix using machine learning algorithms and finite element (FE) analysis. The mass fraction of the ZrO2 particles varied from 4% to 8%, and three sets of solid cylindrical rods with Al4%ZrO2, Al6%ZrO2, and Al8%ZrO2 were prepared using the stir casting method. The upsetting process was simulated, and an equation for predicting hardness was developed from the equivalent strain distributions. Artificial neural networks (ANNs), multilinear regression (MLR) along with equations developed from FE analysis were used to train the model for regression analysis, considering the principal stresses, friction factor, anisotropy ratio, effective strain, and hoop strain as input and the magnitude of hardness as output parameters. Regression analysis reveals that ANN (tri-layer network), XGBoost, and MLR algorithms are the best suitable for the given data sets with a root mean square ( R2) greater than 0.95. XGBoost, ANN (narrow), and SVM are linear and are the most recommendable classifier algorithms for the current investigation. Hardness data from ring compression tests were used to validate the results obtained from the trained models with the test results.

Statistical indicators to evaluate the performance of electrocoagulation with membrane bioreactor (EC-MBR) for treatment of organic matters in domestic wastewater

An electrocoagulation with membrane bioreactor technique (EC-MBR) was developed to treat domestic wastewater and prevent membrane fouling. To support the new design, experiments were conducted on a few levels. The structure and distribution of organic matter removal utilizing the membrane are investigated using a laboratory-scale (EC-MBR) treatment of domestic wastewater. The study's goals were to assess the removal efficiency of organic matter (biological oxygen demand (BOD) and chemical oxygen demand (COD) in Al-Hawraa's wastewater, as well as its links to statistical indicators. It was chosen to sample and evaluate effluent from domestic wastewater using EC-MBR with operating temperature (25 0C), pH (7-8), DO (4-6) mg/L, beginning and final concentrations of BOD (184-6 mg/L), and COD (489-20 mg/L) using biological and electrochemical treatment procedures. According to the results, the organic matter removal efficiency may be calculated using the multilinear regression (MLR) and neural network (NN) models in the SPSS modeler. In addition, the results showed that the entire reactor had good BOD and COD maximum removal efficiencies of 96.7% and 95.9%, respectively. Finally, the highest accuracy of the MLR algorithm for COD and BOD is 99.6 for both, whereas the maximum accuracy of the NN algorithm for COD and BOD is 99.2 % and 99.1%, respectively. To choose the best algorithm for analysis and modeling the outcomes, a comparative study has been achieved to compare the results of two algorithms that used in this study. Therefore, for this study MLR algorithm was chosen.

Satellite based integrated approaches to modelling spatial carbon stock and carbon sequestration potential of different land uses of Northeast India

The study aims to estimate and predict the aboveground biomass, carbon stock and carbon sequestration potential of different land uses of Northeast India and relate these estimates with the land use changes. Many applications such as carbon stock and sequestration monitoring, forest degradation monitoring, and climate change mitigation, require precise and timely estimation of forest biomass. Although traditional field inventory can reliably estimate forest biomass, remote sensing is emerging as an alternate and fast approach to cover larger area with relative precision for biomass estimation. In this study, a combined approach of field inventory and Landsat OLI derived vegetation indices were used in spatial modelling of aboveground biomass and carbon stock in different land uses. A stepwise multilinear regression algorithm was used to derive the model that used Landsat derived NDVI, SAVI and ARVI as predicators. The predicted AGB ranged from 14.32 to 185.95 Mg ha-1 with an average of 148.78 Mg ha-1. The developed model that used combined vegetation indices showed correlation of R2 = 0.79 with an RMSE of 51.04 Mg ha-1. The present study also applied the empirical model (CO2FIX) to simulate the future scenario of carbon stock and carbon sequestration potential of the different land uses. The carbon stock potential of different land uses were 182.31 Mg ha-1, 158.91 Mg ha-1, 134.98 Mg ha-1, 169.26 Mg ha-1, 133.84, 89.95 Mg ha-1, 128.3 Mg ha-1 and 61.36 Mg ha-1 in Tropical forest, Subtropical forest, Temperate forest, Tropical plantation, Subtropical plantation, Temperate plantation, Shifting fallows and Agricultural land, respectively.

A Machine-Learning-Based Labelling Diversity Model for Predictive Analysis: Using 16QAM as a Case Study

The recent advancement and enhancement that optimized uncoded space-time labelling diversity (USTLD) have provided significant diversity gains. By adopting the use of evolutionary algorithms, labelling diversity (LD) mapper designs produced are near-optimal in quality. The only disadvantage to the use of evolutionary algorithms is that the produced solution is not always optimal. To ease the calculation of how much a mapper design had achieved LD, this paper proposes a machine learning-based analysis to predict the amount of LD achieved by a mapper. In this paper, only the 16QAM constellation is studied as a simple case. Six machine learning-based algorithms were proposed in this paper, namely multi-linear regression (MLR), support vector regression (SVR), decision trees (DT), random forest (RF), K-nearest neighbours (KNN) and a simple artificial neural network (ANN). From the results obtained from the experiments, it can be seen that the MLR algorithm is the least time complex while the ANN is the most time complex. It is also important to note that the DT and KNN algorithms take a comparatively short amount of time to execute. When compared in terms of machine learning metrics, it was shown that the ANN algorithm performed the best with the least amount of error while the MLR algorithm performed the worst with the highest amount of error. Thus, it could be seen that the results from this paper provide a positive outlook on applying machine learning algorithms to the LD problem.

Application of multivariate linear regression models for selection of deep eutectic solvent for extraction of apigenin and luteolin from Chrysanthemum indicum L.

Among a variety of compounds presented in chrysanthemum, apigenin and luteolin are the two main components that play a major role in numerous biological activities of this herb. We aimed to obtain linear models showing the dependence of the yield of extraction of apigenin and luteolin on the composition of deep eutectic solvent and investigate the extraction of these two ingredients from Chrysanthemum indicum L. Two models showing the dependence of luteolin and apigenin concentrations on the composition of the solvent were established using a multilinear regression algorithm and were applied to screen 119 different solvents. After that, the extraction process was optimized using response surface methodology and an artificial neural network. Apigenin and luteolin were recovered from the extract by the combination of distillation and addition of water. The screening results on 119 solvents revealed that choline chloride-acetic acid (1:4) was the most suitable deep eutectic solvent. It was showed that both response surface methodology and the artificial neural network could accurately determine the optimal conditions of extraction of apigenin and luteolin from C. indicum L., including time of extraction (65 minutes), temperature of extraction (90°C) and water content (20%). By the combination of distillation and addition of water, apigenin and luteolin could be effectively recovered from the deep eutectic solvent extract with a recovery rate of over 80%. Deep eutectic solvent could be used as an effective green alternative to the conventional solvents for the extraction of bioactive compounds from plants.

Dynamic prediction of energy and power usage cost using linear regression-machine learning analysis

As a known fact, energy usage and demand exponentially rises year after year, hence forth power based companies are apparently looking out for a forecasting approach with better approximations. Based on the usage history at the customer level with the emergence of machine learning, and its association with various prediction and decision making fields. This paper aims to use a machine learning algorithm to predict the cost levied on the customer proportional to the usage. The efficacy of this model is compared to the results obtained with the mathematical computations. It is evident that the accuracy is 95% with reference to the multilinear regression algorithm

A Virtual Geostationary Ocean Color Sensor to Analyze the Coastal Optical Variability

In the coastal environment the optical properties can vary on temporal scales that are shorter than the near-polar orbiting satellite temporal resolution (~1 image per day), which does not allow capturing most of the coastal optical variability. The objective of this work is to fill the gap between the near-polar orbiting and geostationary sensor temporal resolutions, as the latter sensors provide multiple images of the same basin during the same day. To do that, a Level 3 hyper-temporal analysis-ready Ocean Color (OC) dataset, named Virtual Geostationary Ocean Color Sensor (VGOCS), has been created. This dataset contains the observations acquired over the North Adriatic Sea by the currently functioning near-polar orbiting sensors, allowing approaching the geostationary sensor temporal resolution. The problem in using data from different sensors is that they are characterized by different uncertainty sources that can introduce artifacts between different satellite images. Hence, the sensors have different spatial and spectral resolutions, their calibration procedures can have different accuracies, and their Level 2 data can be retrieved using different processing chains. Such differences were reduced here by adjusting the satellite data with a multi-linear regression algorithm that exploits the Fiducial Reference Measurements data stream of the AERONET-OC water-leaving radiance acquired at the Acqua Alta Oceanographic Tower, located in the Gulf of Venice. This work aims to prove the suitability of VGOCS in analyzing the coastal optical variability, presenting the improvement brought by the adjustment on the quality of the satellite data, the VGOCS spatial and temporal coverage, and the inter-sensor differences. Hence, the adjustment will strongly increase the agreement between the satellite and in situ data and between data from different near-polar orbiting OC imagers; moreover, the adjustment will make available data traditionally masked in the standard processing chains, increasing the VGOCS spatial and temporal coverage, fundamental to analyze the coastal optical variability. Finally, the fulfillment by VGOCS of the three conditions for a hyper-temporal dataset will be demonstrated in this work.

Characteristics of Black Carbon Particle-Bound Polycyclic Aromatic Hydrocarbons in Two Sites of Nanjing and Shanghai, China

Airborne polycyclic aromatic hydrocarbons (PAHs) are of great concern to human health due to their potential high toxicity. Understanding the characteristics and sources of PAHs, as well as the governing factors, is therefore critical. PAHs and refractory black carbon (rBC) are both from combustion sources. This work, for the first time, investigated exclusively the rBC-bound PAH properties by using a laser-only Aerodyne soot-particle aerosol mass spectrometer (SP-AMS). This technique offers highly time-resolved PAH results that a traditional offline measurement is unable to provide. We analyzed two datasets conducted in urban Shanghai during the fall of 2018 and in suburban Nanjing during the winter of 2017, respectively. Results show that the average concentration of PAHs in Nanjing was much higher than that in Shanghai. Nanjing PAHs contained more low molecular weight components while Shanghai PAHs contained more high molecular weight ones. PAHs in Shanghai presented two peaks in early morning and evening, while Nanjing PAHs had only one significant morning peak, but remained high throughout the nighttime. A multi-linear regression algorithm combined with positive matrix factorization (PMF) analyses on sources of PAHs reveals that the industry emissions contributed the majority of PAHs in Nanjing (~80%), while traffic emissions dominated PAHs in Shanghai (~70%). We further investigated the relationships between PAHs with various factors. PAHs in both sites tended to positively correlate with primary pollutants, including primary organic aerosol (OA) factors, and gaseous pollutants of CO, NO2 and SO2, but negatively correlated with secondary OA factors and O3. This result highlights the enhancement of rBC-bound PAHs level due to primary emissions and their oxidation loss upon atmospheric aging reactions. High concentration of PAHs seemed to frequently appear under low temperature and high relative humidity conditions, especially in Shanghai.

Diabetics Prediction Based on Multi-Linear Regression Using R Language

Classification is an important technique in data mining which is applied in many fields including medical diagnosis to find diseases. In this research work, the multi-linear regression algorithm is used to find the possibilities of occurrence of diabetes. This research work would help the developers to identify the characteristics and flow of algorithms. This implementation helps the diabetalogist to make decision quickly. The explict outcome of the performance of the algorithm is reported for the chosen data.

Empirical Modeling of the Equatorward Boundary of Auroral Precipitation Using DMSP and DE 2

AbstractA new model of the equatorward boundary of the diffuse aurora has been developed using observations of precipitating particles made by the Defense Meteorological Satellite Program (DMSP) from 1987 to 2012 as well as Dynamics Explorer 2 (DE 2), which operated from August 1981 to February 1983. Using a local multilinear regression algorithm, we investigated the use of different combinations of magnetometer indices and solar wind coupling functions with different averaging periods and weights to find the best parameterization for a model of the equatorward boundary of the aurora. We find that weighted averages of the AE index and the solar wind coupling function dΦMP/dt both outperform the often used Kp index. Using conjunction events where two DMSP satellites cross the auroral boundary at nearly the same time, we show that these models are better at extrapolating the auroral boundary to different local times than previous models.

Ultra-Short-Term Wind Power Prediction Based on Multivariate Phase Space Reconstruction and Multivariate Linear Regression

In order to improve the accuracy of wind power prediction (WPP), we propose a WPP based on multivariate phase space reconstruction (MPSR) and multivariate linear regression (MLR). Firstly, the multivariate time series (TS) are constructed through reasonable selection of wind power and weather factors, which are closely associated with wind power. Secondly, the phase space of the multivariate time series is reconstructed based on the chaos theory and C-C method. Thirdly, an auto regression model for multivariate phase space is created by regarding phase variables as state variables, and the very-short-term wind power is predicted by using a multi-linear regression algorithm. Finally, a parallel algorithm based on map/reduce is presented to improve computing speed. A cloud computing platform, Hadoop consisting of five nodes, is established as a matter of convenience, followed by the prediction of wind power of a wind farm in the Hunan province of China. The experimental results show that the model based on MPSR and MLR is more accurate than both the continuous method and the simple approximation method, and the parallel algorithm based on map/reduce effectively accelerates the computing speed.

부분최소자승법을 이용한 중고차 에어컨냄새 원인물질 추정

Lately, customers think highly of the emotional satisfaction and as a result, issues on odor are matters of concern. The cases are odor of interior material and air-conditioner of vehicles. In particualar, with respect to the odor of air-conditioner, customers strongly claimed defects with provocative comments : "It smells like something rotten," "It smells like a foot odor," "It stinks like a rag." Generally, it is known that mold of evaporator core in the air-conditioning system decays and this produce VOCs which causes the odor to occur. In this study, partial least squares regression model is applied to predict the strength of the odor and select of important VOCs which affect car air conditioning smell. The PLS method is basically a particular multilinear regression algorithm which can handle correlated inputs and limited data. The number of latent variable is determined by the point which is stabilized mean absolute deviations of VOCs data. Also multiple linear regression is carried out to confirm the validity of PLS method.

Total alkalinity estimation using MLR and neural network techniques

During the last decade, two important collections of carbon relevant hydrochemical data have become available: GLODAP and CARINA. These collections comprise a synthesis of bottle data for all ocean depths from many cruises collected over several decades. For a majority of the cruises at least two carbon parameters were measured. However, for a large number of stations, samples or even cruises, the carbonate system is under-determined (i.e., only one or no carbonate parameter was measured) resulting in data gaps for the carbonate system in these collections. A method for filling these gaps would be very useful, as it would help with estimations of the anthropogenic carbon (Cant) content or quantification of oceanic acidification. The aim of this work is to apply and describe, a 3D moving window multilinear regression algorithm (MLR) to fill gaps in total alkalinity (AT) of the CARINA and GLODAP data collections for the Atlantic. In addition to filling data gaps, the estimated AT values derived from the MLR are useful in quality control of the measurements of the carbonate system, as they can aid in the identification of outliers. For comparison, a neural network algorithm able to perform non-linear predictions was also designed. The goal here was to design an alternative approach to accomplish the same task of filling AT gaps. Both methods return internally consistent results, thereby giving confidence in our approach.

Quantitative structure–activity relationship modeling of bioconcentration factors of polychlorinated biphenyls

The bioconcentration factors (BCFs) of 57 polychlorinated biphenyl (PCB) congeners were modeled by quantitative structure–activity relationship (QSAR) based on 486 constitutional, topological, geometrical, electrostatic, quantum chemical, and thermodynamic descriptors derived solely from molecular structure and calculated using CODESSA Pro (comprehensive descriptors for structural and statistical analysis) software. Descriptors utilized for the general model were selected by various statistical validation techniques. Multilinear models were developed using the best multilinear regression algorithm to relate experimental BCF to a set of molecular descriptors. The proposed two-parameter model satisfactorily describes the relationship between observed and calculated values in terms of statistical parameters. Polarity, structural flexibility, and spatial mass distribution of a molecule were demonstrated to be the main factors influencing the ability of PCB to (1) penetrate through lipophilic cell membranes and (2) bind specifically and nonspecifically to biological targets, hence affecting BCF. Comparison to other models indicated advantages of the proposed model over previously reported ones. Derived two-parameter regression equation has improved statistics and is based on theoretical descriptors with a definite physicochemical meaning; it is easier to use and interpret due to the mathematical simplicity of the linear QSAR approach. Internal validation and scrambling procedure confirmed the stability and reliable predictive ability of the general model and indicated the absence of chance correlations. External validation demonstrated that the presented model can be applied to structurally similar sets of compounds, thus extending the domain of applicability of the model.

Quantitative Structure–Activity Relationship (QSAR) Modeling of EC50 of Aquatic Toxicities for Daphnia magna

The experimental EC50 toxicities toward Daphnia magna for a series of 130 benzoic acids, benzaldehydes, phenylsulfonyl acetates, cycloalkane-carboxylates, benzanilides, and other esters were studied using the Best multilinear regression algorithm (BMLR) implemented in CODESSA. A modified quantitative structure–activity relationships (QSAR) procedure was applied guaranteeing the stability and reproducibility of the results. Separating the initial data set into training and test subsets generated three independent models with an average R 2 of .735. A five-descriptor general model including all 130 compounds, constructed using the descriptors found effective for the independent subsets, was characterized by the following statistical parameters: R 2 = .712; R 2 cv = .676; F = 61.331; s2 = 0.6. The removal of two extreme outliers improved significantly the statistical parameters: R 2 = .759; R2 cv = .728; F = 77.032; s2 = 0.499. The sensitivity of the general model to chance correlations was estimated by applying a scrambling procedure involving 20 randomizations of the original property values. The resulting R 2 = .192 demonstrated the high robustness of the model proposed. The descriptors appearing in the obtained models are related to the biochemical nature of the adverse effects. An additional study of the EC50/LC50 relationship for a series of 28 compounds (part of our general data set) revealed that these endpoints correlated with R 2 = .98.

A Statistical Method for Generating Cross-Mission Consistent Normalized Water-Leaving Radiances

The accurate merging of primary radiometric ocean color products such as the normalized water-leaving radiance requires combining data from various space missions, which may be affected by different uncertainties as resulting from absolute calibration and minimization of the atmospheric effects. A statistical correction scheme based on a multilinear regression algorithm is used here to remove systematic differences between in situ and remote-sensing measurements. The application of the correction scheme to Sea-viewing Wide Field-of-View Sensor (SeaWiFS) and Moderate Resolution Imaging Spectroradiometer (MODIS) primary radiometric products improves the convergence between remote-sensing and in situ measurements, with the largest effects at 412 and 443 nm. Specifically, the scatter and bias of MODIS derived with respect to in situ L wn at 412 nm have shown values of 12% and 3% for corrected with respect to values of 34% and -28% for uncorrected data, respectively. Similarly, the scatter and bias for SeaWiFS-derived L wn at 412 nm have shown values of 14% and 4% for corrected with respect to 32% and -20% for uncorrected data. Results at 667 nm for MODIS and at 670 nm for SeaWiFS, although displaying a reduction in the scatter of data, have shown a significant residual bias of about 11% and 17% with respect to in situ values. Finally, it was shown the need for restricting the application of the correction scheme to data with atmospheric and marine optical features represented within the reference data set used to define the correction coefficients.

Near-infrared (NIR) analysis of day 2 single embryo transfer culture media samples: blind validation of a multi-linear regression algorithm

Near-infrared (NIR) analysis of day 2 single embryo transfer culture media samples: blind validation of a multi-linear regression algorithm

A comparison of Multi-layer Perceptron and multilinear regression algorithms for the inversion of synthetic ocean colour spectra

Artificial radiance sets were used as inputs to Multi-layer Perceptron and multilinear regression algorithms to study their retrieval capabilities for optically active constituents in sea water. The radiative transfer model Hydrolight was used to produce 18,000 artificial reflectance spectra representing various case 1 and case 2 water conditions. The remote sensing reflectances were generated at the Medium Resolution Imaging Spectrometer (MERIS) wavebands 412, 442, 490, 510, 560, 620, 665 and 682 nm from randomly generated triplet combinations of chlorophyll a, non-chlorophyllous particles and CDOM (Coloured Dissolved Organic Matter) concentrations. These reflectances were contaminated with different noise terms, before they were used to assess the performance of multilayer perceptron and multilinear regression algorithms. The potential of both algorithms for retrieving optically active constituents was demonstrated with the neural network showing more accurate results for case 2 scenarios.

Multicomponent sequential injection analysis determination of nitro-phenols in waters by on-line liquid-liquid extraction and preconcentration

An automatic sequential injection analysis (SIA) set-up for the determination of phenolic compounds (2-, 3- and 4-nitrophenol) in waters was designed and evaluated. The system performs an on-line phenol extraction from an aqueous medium to an organic solvent, which enables the analyte isolation and preconcentration. The extracted phenols are on-line back-extracted into a small volume of NaOH solution and conducted to a spectrophotometric detector for quantitative analysis. This methodology is based on the formation of a thin organic solvent layer adhered to the PTFE surface of the extraction coil. In the present work, no reversed-flow was carried out in both the extraction and back-extraction steps. By using a photodiode-array detector, multilinear regression algorithms and first derivative spectroscopic techniques were evaluated in order to perform multicomponent determinations of phenol mixtures. Concentrations within the range 0.4–4.6 mg l −1 of 2-nitrophenol, 0.6–11.2 mg l −1 of 3-nitrophenol and 55.6–938 mg l −1 of 4-nitrophenol have been determined and simultaneously analysed at a frequency of four samples per hour. A R.S.D. of 2.2% was obtained for 2.6 mg l −1 of 2-nitrophenol, 2.3% for 5.3 mg l −1 of 3-nitrophenol and 2.2% for 330.8 μg l −1 of 4-nitrophenol, with preconcentration factors comprised between seven and nine.

Molecular mechanical parameters (MM2 force field) for the N(sp 3)-O(sp 3) bond

Based on results of MP2/6-31G* ab initio calculations an MM2 molecular mechanical parameter set has been developed for molecules containing N(sp3)-O(sp3) single bonds, existing parameters concerning the other bonds being retained. The new parameter set was tested for small organic compounds. A simple, generally applicable multilinear regression algorithm has been used and a program written to complement an existing force field (e.g. MM2) with such parameters extracted from quantum chemical computations.