Multiple Polynomial Regression Research Articles

Biosorption of pb(II) from aqueous solution by citrus reticulate: adsorption studies, and modeling

Removing toxic Pb(II) ions from aqueous solution by the peels of citrus reticulate (mandarin orange), a fruit industry waste, presents suitable scale-up possibilities. The Scanning Electron Microscope (SEM) and Brunauer-Emmett-Teller (BET) studies reflected that the mandarin orange peel powder had a porous surface area (32.46 m2g−1), average pore size and pore volume was 38.6 Å and 0.402 cm3g−1, respectively, favorable for binding Pb(II) ions. Fourier-transform infrared spectroscopy (FTIR) showed C-Br stretching, primary alcohol (C-O), phenolic O-H, and carbodimide N = C = N bands primarily helped to bind Pb(II) ions. The study evaluated and optimized the parametric influences of pH, adsorbate and biosorbent concentration, contact time and temperature on the removal efficiency of Pb(II) ions. A maximum of 97.08% Pb(II) was removed from 20 mg L−1 solution when 2.5 g L−1 adsorbent was present. The reaction obeyed the pseudo-second-order kinetic model. The intra-particle diffusion was involved in lead sorption. The Langmuir isotherm model resulted in an adsorption capacity of 23.04 mg g−1. 35.28% Pb(II) was removed in the 3rd adsorption-desorption cycle with 0.4 M HCl. The adsorption process was natural, impulsive and endothermic. The statistical investigation used Multiple Polynomial Regression (MPR) and Genetic Algorithm (GA). The analysis effectively forecasted the percentage removal at the optimized condition.

What makes for digital entrepreneurs? The role of AI-related drivers for nascent digital start-up activities

PurposeApplying the Stimulus–Organism–Response (SOR) model, this study aims to explore how AI-driven stimuli (e.g. ChatGPT adoption in entrepreneurship and perceived AI competencies) stimulate individuals’ cognitive organisms (e.g. digital entrepreneurial opportunity exploration and exploitation), and how these individually, congruently, and incongruently trigger their behavioral responses (e.g. nascent digital start-up activities).Design/methodology/approachUtilizing a sample of 1326 MBA students in Vietnam with a stratified sampling approach, multiple linear regression and polynomial regression with response surface analysis were used to test hypotheses.FindingsThe findings reveal that ChatGPT adoption in entrepreneurship and perceived AI competencies have a positive and significant impact on individuals’ digital entrepreneurial opportunity exploration and exploitation, which in turn, positively affects nascent digital start-up activities. Moreover, the study also reports that digital entrepreneurial opportunity exploration and exploitation can be congruently combined with each other to trigger the effects of nascent digital start-up activities.Practical implicationsSome valuable recommendations based on the findings have been provided for practitioners and policymakers.Originality/valueThe study contributes to the academic landscape by validating the SOR model within the context of AI adoption and entrepreneurship. It emphasizes the sequential processes of stimulus, cognitive responses, and behavioral outcomes, shedding light on nuanced effects in the digital entrepreneurial landscape.

Retraining surrogate models in increasingly restricted design spaces: a novel building energy model calibration method

Surrogate (i.e. meta) models can approximate building energy models (BEMs) accurately and quickly, hence they have been widely used in BEM calibration studies. Typically, the surrogate models are trained a single time over the entire unknown building parameter space with a design such as Latin hypercube sampling. In this article, a multiple polynomial regression surrogate model is, instead, retrained with increasingly restricted designs. In each training repetition, the bounds of the design narrow around the unknown building parameter values that minimize the error between the surrogate model’s predictions and the measured energy. This ‘cascading surrogate’ calibration method finds CVRMSE values that are much lower than those of a powerful black box optimizer in a case study with simulated ‘measured’ data. However, the method has similar performance to the black box optimizer in a case study with real hourly measured energy, probably since the BEM was not configured accurately enough.

Optimization of Mg-Al Layered Double Hydroxide Film Preparation and Corrosion Resistance Study on AZ91D Mg Alloy by Multivariate Polynomial Regression Fitting.

Layered double hydroxide (LDH) films have received extensive attention for their unique physical barrier function and ion exchange properties, which make them promising candidates for corrosion protection of magnesium alloys. In this paper, we used the multiple polynomial regression fitting method to establish a regression equation for the electrochemical corrosion resistance with the reaction temperature (T), pH, and reaction time (t) of the Mg-Al LDH film on the AZ91D magnesium alloy. The goodness of fit, confidence, and residual analyses confirmed the high accuracy of the model equation. According to the calculation using the fmincon function, the best corrosion resistance of the prepared samples could be achieved when the parameters are T = 135 °C, pH = 12.0, and t = 15 h. Then, the experimental results showed that the corrosion current density (Icorr) of the obtained LDH film under the above conditions could be 1.07 × 10-7 A/cm2, approximately 3 orders of magnitude lower than the magnesium alloy substrate, after immersion in a 3.5 wt % NaCl solution for 180 h, the surface structure of the LDH film did not change significantly, and the Icorr was still 2 orders of magnitude higher than that of the magnesium alloy substrate. Hence, a synergistic effect equation for the reaction temperature, pH, and reaction time on the corrosion resistance of the LDH film on a magnesium alloy surface prepared by the hydrothermal method was obtained. Moreover, using this equation, we obtained an LDH film with good corrosion resistance and durability, providing theoretical guidance for optimizing the process of preparing the LDH film by the hydrothermal method in practical applications.

Data-driven prediction of the surface layer state in hard-turning for optimization of component quality

Data-driven approaches are an effective solution for modeling problems in machining. To increase the service life of hard-turned components, it is important to quantify the correlation between the cutting parameters such as feed rate, cutting speed and depth of cut and the near-surface properties. For obtaining high-quality models with small data sets, different data-driven approaches are investigated in this contribution. Additionally, models that enable uncertainty quantification are crucial for effective decision-making and the adjustment of cutting parameters. Therefore, parametric multiple polynomial regression and Takagi–Sugeno models, as well as non-parametric Gaussian process regression as a Bayesian approach are considered and compared regarding their capability to predict residual stress and surface roughness values of 51CrV4 specimens after hard-turning. Moreover, a novel method based on optimization of data driven non-linear models is proposed that allows for identification of cutting parameter combinations, which at the same time lead to satisfactory surface roughness and residual stress states.

Modeling of Capacity Factor in Rembang Coal-Fired Steam Power Plant Using Regression Modelin

Coal-Fired Steam Power Plant (PLTU) Rembang is an important power plant in the Central Java electricity system. Like other coal-fired steam power plants, fuel cost is the most significant expense when operating the PLTU Rembang. During the 2019-2021 period, the average fuel cost was 73.88% of total costs. One of the ways to reduce fuel costs is by improving the accuracy of fuel demand planning. Fuel procurement planning is very dependent on the projected amount of electricity sales from power plant, which is largely determined by the power plant's Capacity Factor (CF). However, PLTU Rembang does not have any CF prediction modeling. This research developed and compared four prediction models: random forest regression, support vector regression, multiple polynomial regression, and multiple linear regression. Based on the comparison of validation from the four prediction model with MAPE and R-squared parameters, the multiple linear regression models is the best model, with the lowest MAPE of 7.83% and the highest R-squared of 0.8814. This multiple linear regression model can be used to predict the CF of PLTU Rembang in the future so that fuel demand planning is more accurate.

Multiple Polynomial Regression Model for Predicting Surface Roughness of Titanium Alloy in Electrical Discharge Machining

Multiple Polynomial Regression Model for Predicting Surface Roughness of Titanium Alloy in Electrical Discharge Machining

Application of Artificial Intelligence in Predicting Oil Production Based on Water Injection Rate

The utilization of artificial intelligence (AI) has become imperative across various domains, including the oil and gas industry, which covers several fields, including reservoirs, drilling, and production. In oil and gas production, conventional methods, such as reservoir simulation, are used to predict the oil production rate. This simulation requires comprehensive data, so each process step takes a long time and is expensive. AI is urgently needed and can be a solution in this case. This research aims to apply AI techniques to forecast oil production rates based on water injection rates from two injection wells. Three wells are connected with a direct line drive pattern. Three different AI methods were applied, including multiple linear polynomial regression (PR), multiple linear regression (MLR), and artificial neural networks (ANN) in constructing oil production rate prediction models. Actual field data of 1180 data are used, including water injection rate data from two injection wells and oil production history data from one production well. The dataset has been split randomly into 80% training and 20% allocated for testing subsets. The training data is used to build predictive models, while the testing data is used to validate model performance. Comparative analysis selects the model with the lowest root mean square error (RMSE) and the highest R^2 test value. Results demonstrate that the ANN model achieves the smallest Root Mean Square Error (RMSE) of 0.142 and the highest R^2 test value of 16.2%, outperforming the PR and MLR methods. The ANN prediction model provides a rapid and efficient approach to estimating oil production rates.

Application of Artificial Intelligence in Predicting Oil Production Based on Water Injection Rate

The utilization of artificial intelligence (AI) has become imperative across various domains, including the oil and gas industry, which covers several fields, including reservoirs, drilling, and production. In oil and gas production, conventional methods, such as reservoir simulation, are used to predict the oil production rate. This simulation requires comprehensive data, so each process step takes a long time and is expensive. AI is urgently needed and can be a solution in this case. This research aims to apply AI techniques to forecast oil production rates based on water injection rates from two injection wells. Three wells are connected with a direct line drive pattern. Three different AI methods were applied, including multiple linear polynomial regression (PR), multiple linear regression (MLR), and artificial neural networks (ANN) in constructing oil production rate prediction models. Actual field data of 1180 data are used, including water injection rate data from two injection wells and oil production history data from one production well. The dataset has been split randomly into 80% training and 20% allocated for testing subsets. The training data is used to build predictive models, while the testing data is used to validate model performance. Comparative analysis selects the model with the lowest root mean square error (RMSE) and the highest R^2 test value. Results demonstrate that the ANN model achieves the smallest Root Mean Square Error (RMSE) of 0.142 and the highest R^2 test value of 16.2%, outperforming the PR and MLR methods. The ANN prediction model provides a rapid and efficient approach to estimating oil production rates.

Pico scale undershot waterwheel for ultra-low-head: Analytical, experimental and CFD method

In the 19th and early 20th century, the utilization and development of the undershot waterwheel (USWW) was massif. USWW represents a global warming and crisis electricity solution issue for the world. However, the best condition and design of USWW are still incomplete; previous research is based on assumptions. Hence, this work aims to theoretically estimate the power loss (head loss, leakage loss, and friction loss) occurring in USWW to predict the relation between the blade number and optimum operation. The investigation employs analytical and CFD methods, then validated with experimental data. Then, based on root mean square error (RMSE) analysis, the average RMSE of analytical results to experimental data was 9.02%, and of CFD data was 9.07%. Then, depict the blade number and optimum operation relation to USWW performance expressing using multiple quadratic polynomial regression. The empirical law identifies the 16-blades, and the working operation of 0.5 is USWW maximum performance in ultra-low-head.

Leachate Treatment via TiO2/UV Heterogeneous Photocatalysis: A Multiple Polynomial Regression Model

Advanced oxidation processes such as TiO2/UV heterogeneous photocatalysis are suitable treatment methods for wastewater with high pollutant loads such as landfill leachates. Optimizing the variables that influence the process is a fundamental aspect. However, in this regard, experimental conditions are limited in terms of resources and time, which is why modeling allows obtaining a general understanding of the phenomenon from a set of experimental data. This work sought to model the photocatalytic process via multivariate polynomial regression, considering variables such as the catalyst concentration, the pH level, and the accumulated energy concerning the percentage of degradation in terms of dissolved organic carbon (DOC). The implemented fitting method resulted in a third-degree polynomial with an R2 of 0,8652, concluding that the model and its conclusions are valid. Moreover, with greater degrees, the model curve overfitted, even with better R2. DOC abatement showed a negative correlation with pH and the catalyst dose, while an opposite trend was observed for the accumulated energy. The model predictions allow inferring that, at low catalyst doses and medium and high pH levels, it is possible to find maximum degradations at low cumulative energies.

Multiple Stock Prediction Based on Linear and Non-linear Machine Learning Regression Methods

In contemporary times, the pressing issue of global environmental pollution has prompted the exploration of alternative energy sources by various industries, aiming to mitigate the adverse environmental impacts caused by traditional energy production. Correspondingly, investors in the financial market have increasingly redirected their capital towards the new energy sector. Within this context, the present research endeavors to employ machine learning techniques for the prediction of Tesla's stock price. This study leverages multiple linear regression, polynomial regression, and lag models to construct models based on the datasets of TSLA, MPC, and UNG stock prices spanning the period of 2019-2020. By discerning potential patterns among these variables, the objective is to anticipate the future trajectory of TSLA stock price. According to machine learning methods, Tesla's stock price can be predicted, and the daily price of Tesla is influenced by the opening price, high price, low price and trading volume of the stock on that day. In addition, the share prices of energy companies related to Tesla also have an impact on Tesla's share price on that day. Specifically, Tesla's stock price is influenced by Natural Gas Company (UNG), which has an opposite relationship. Although common sense economics says that the crude oil market will be closely related to the new energy market. However, the results of this study demonstrated that Tesla's stock price is less influenced by Crude Oil Company (MPC).

Fuzzy Modeling Development for Lettuce Plants Irrigated with Magnetically Treated Water.

Due to the worldwide water supply crisis, sustainable strategies are required for a better use of this resource. The use of magnetic water has been shown to have potential for improving irrigation efficacy. However, a lack of modelling methods that correspond to the experimental results and minimize error is observed. This study aimed to estimate the replacement rates of magnetic water provided by irrigation for lettuce production using a mathematical model based on fuzzy logic and to compare multiple polynomial regression analysis and the fuzzy model. A greenhouse study was conducted with lettuce using two types of water, magnetic water (MW) and conventional water (CW), and five irrigation levels (25, 50, 75, 100 and 125%) of crop evapotranspiration. Plant samples for biometric lettuce were taken at 14, 21, 28 and 35 days after transplanting. The data were analyzed via multiple polynomial regression and fuzzy mathematical modeling, followed by an inference of the models and a comparison between the methods. The highest biometric values for lettuce were observed when irrigated with MW during the different phenological stage evaluated. The fuzzy model provided a more exact adjustment when compared to the multiple polynomial regressions.

Adsorptive elimination of methylene blue dye from aqueous solution by chitosan‐nSiO2 nanocomposite: Adsorption and desorption study, scale‐up design, statistical, and genetic algorithm modeling

AbstractHeavy metals and organic dyes in water in concentrations exceeding the tolerable limit are unsafe for aquatic life and humans. Chitosan nanocomposites show a potential adsorption capacity to organic dyes. The present study encompasses the preparation of chitosan‐nSiO2 nanocomposites (CSNC) with different weight ratios of chitosan to nSiO2 and its application in the methylene blue (MB) dye adsorption. The nanocomposites were characterized with the help of SEM, BET, FTIR, XRD, and TGA. The adsorption experimentation was executed in batch mode under varying experimental conditions. Several isotherm and kinetic models were analyzed with the experimental data, and thermodynamic conditions required for adsorption were also determined. Maximum Langmuir adsorption capacities (qL) of the adsorbents for MB varied in the range of 21.32–31.34 mg g−1. The pseudo‐second‐order model was the best‐fitted kinetic model concerning all three adsorbents. The statistical modeling using multiple polynomial regression (MPR) for CWS CSNC1‐1 and CSNC2‐1 yielded the equations with R‐square ranging from 0.984 to 0.996. For GA modeling, it is more than 0.999. So, the efficient employment of statistical modeling, and genetic algorithms has also been achieved.

Enhanced adsorption performance of tetracycline in aqueous solutions using Mg-Al-LDH/AC nanocomposite

Water contamination with antibiotics promotes environmental antibiotic resistance. Tetracycline(TC), commonly used in agriculture, pollutes water and enables bacteria to develop resistance genes. Mg-Al-layered double hydroxides/Activated carbon (Mg-Al-LDH/AC) nanocomposite effectively removes tetracycline from water in this study. The synthesized adsorbent was characterized using Brunauer–Emmett–Teller (BET), Fourier-transform Infrared Spectroscopy (FT-IR), Field Emission Scanning Electron Microscopy (FESEM), X-Ray Diffraction (XRD), and Energy-dispersive X-ray spectroscopy (EDS) tests, and exhibited a large surface area of 105.46 m2/g, a total pore volume of 0.4571 cm3/g, and an average pore diameter of 27.99 nm. The optimal conditions for eliminating TC using the Mg-Al-LDH/AC nanocomposite were determined via the central composite design response surface methodology (CCD-RSM) approach. The responses were evaluated using a second-order polynomial multiple regression model, which was found to be a satisfactory fit to the data based on the analysis of variance (R2 = 0.9690, R2Adj = 0.9478, and R2Pred = 0.9042). The ideal conditions were identified as pH = 5.1, 0.11 g of adsorbent, a TC concentration of 50 mg/L, and a contact time of 90 min. The adsorption isotherms showed that TC elimination follows the Langmuir model with a maximum adsorption capacity of 106.4 mg/g and a high R2 value of 0.9986. The kinetics of the TC adsorption process was found to fit the pseudo-second-order model with a correlation value of 0.9997. The values of ΔG, ΔH, and ΔS for TC were determined to be 1.90 kJ/mol (at 333.15 K), −41.97 kJ/mol, and −0.131 kJ/(mol·K), respectively. Additionally, the study showed that ionic strength did not affect the effectiveness of TC removal, and the adsorbent was effective in removing TC antibiotics in a range of ionic strengths. Finally, the study found that adding 20% by weight of AC to LDH resulted in the best TC removal efficiency.

Development of a Machine Learning Model to Estimate the Biotic Ligand Model-Based Predicted No-Effect Concentrations for Copper in Freshwater.

The copper (Cu) biotic ligand model (BLM) has been used for ecological risk assessment by taking into account the bioavailability of Cu in freshwater. The Cu BLM requires data for many water chemistry variables, such as pH, major cations, and dissolved organic carbon, which can be difficult to obtain from water quality monitoring programs. To develop an optimized predicted no-effect concentration (PNEC) estimation model based on an available monitoring dataset, we proposed an initial model that considers all BLM variables, a second model that requires variables excluding alkalinity, and a third model using electrical conductivity as a surrogate for the major cations and alkalinity. Furthermore, deep neural network (DNN) models have been used to predict the nonlinear relationships between the PNEC (outcome variable) and the required input variables (explanatory variables). The predictive capacity of DNN models was compared with the results of other existing PNEC estimation tools using a look-up table and multiple linear and multivariate polynomial regression methods. Three DNN models, using different input variables, provided better predictions of the Cu PNECs compared with the existing tools for the following four test datasets: Korean, United States, Swedish, and Belgian freshwaters. Consequently, it is expected that Cu BLM-based risk assessment can be applied to various monitoring datasets, and that the most applicable model among the three different types of DNN models could be selected according to data availability for a given monitoring database. Environ Toxicol Chem 2023;00:1-13. © 2023 SETAC.

Predicting spatiotemporal soil organic carbon responses to management using EPIC-IIASA meta-models

The management of Soil Organic Carbon (SOC) is a critical component of both nature-based solutions for climate change mitigation and global food security. Agriculture has contributed substantially to a reduction in global SOC through cultivation, thus there has been renewed focus on management practices which minimize SOC losses and increase SOC gain as pathways towards maintaining healthy soils and reducing net greenhouse gas emissions. Mechanistic models are frequently used to aid in identifying these pathways due to their scalability and cost-effectiveness. Yet, they are often computationally costly and rely on input data that are often only available at coarse spatial resolutions. Herein, we build statistical meta-models of a multifactorial crop model in order to both (a) obtain a simplified model response and (b) explore the biophysical determinants of SOC responses to management and the geospatial heterogeneity of SOC dynamics across Europe. Using 5600 unique simulations of crop growth from the gridded Environmental Policy Integrated Climate-based Gridded Agricultural Model (EPIC-IIASA GAM) covering 86,000 simulation units across Europe, we build multiple polynomial regression ensemble meta-models for unique combinations of climate and soil across Europe in order to predict SOC responses to varying management intensities. We find that our biophysically-explicit meta models are highly accurate (R2 = 0.97) representations of the full mechanistic model and can be used in lieu of the full EPIC-IIASA GAM model for the estimation of SOC responses to cropland management. Model stratification by means of climate and soil clustering improved the performance of the meta-models compared to the full EU-scale model. In regional and local validations of the meta-model predictions, we find that the meta-models largely capture broad SOC dynamics such as the linear nature of SOC responses to residue application, yet they often underestimate the magnitude of SOC responses to management. Furthermore, we find notable differences between the results from the biophysically-specific models throughout Europe, which point to spatially-distinct SOC responses to management choices such as nitrogen fertilizer application rates and residue retention that illustrate the potential for these models to be used for future management applications. While more accurate input data, calibration, and validation will be needed to accurately predict SOC change, we demonstrate the use of our meta-models for biophysical cluster and field study scale analyses of broad SOC dynamics with basically zero fine-tuning of the models needed. This work provides a framework for simplifying large-scale agricultural models and identifies the opportunities for using these meta-models for assessing SOC responses to management at a variety of scales.

Modeling and optimizing concentration of exogenous application of γ-aminobutyric acid on NaCl-stressed pineapple mint (Mentha suaveolens) using response surface methodology: an investigation into secondary metabolites and physiological parameters

Salinity, a severe worldwide issue, compromises the economic production of medicinal plants including mints and causes drug-yield decline. γ-Aminobutyric acid (GABA) is a tolerance-inducing signaling bio-molecule in various plant physiological processes. Pineapple mint (Mentha suaveolens Ehrh.) is a valuable medicinal herb with an exhilarating scent of citrus fruit. Piperitenone oxide is the major bioactive constituent of its essential oil, having significant demand by pharmaceutical industries. Nonetheless, modeling and optimizing the effective concentration of GABA remain within twin foci of interest. Therefore, a two factor-five level (NaCl 0-150 mM and GABA 0-2.4 mM) central composite design was conducted to model and optimize drug yield and physiological responses of M. suaveolens. Based on the design of experiments (DoE) approach, different linear, quadratic, cubic, and quartic models were assigned to the response variables. Change trends of shoot and root dry weights followed a simple linear model, whereas sophisticated models (i.e., multiple polynomial regression) were fitted to the other traits. NaCl stress inevitably reduced root and shoot dry weight, piperitenone oxide content, relative water content, pigments content, and maximum quantum yield of PSII. However, content of malondialdehyde (MDA) and total flavonoid, and DPPH radical scavenging activity were increased under salinity. Under severe NaCl stress (150 mM), the essential oil content (0.53%) was increased three times in comparison with control (0.18%). Optimization analysis demonstrated that the highest amount of essential oil (0.6%) and piperitenone oxide (81%) as a drug yield-determining component would be achievable by application of 0.1–0.2 mM GABA under 100 mM NaCl. The highest dry weight of root and shoot was predicted to be achieved at 2.4 mM GABA. Overall, extremely severe NaCl stress (i.e., more than 100 mM) in which a sharp drop in yield components value was observed seemed to be out of M. suaveolens salinity tolerance range. Hence, it is rationale to compensate the decrease of drug yield by foliar application of a dilute GABA solution (i.e., 0.1–0.2 mM) under 100 mM NaCl stress or lower levels.

Predict the thyroid abnormality particular disease likelihood of the symptoms’ certainty factor value and its confidence level: A regression model analysis

The traditional expert system (TES) in the medical field commonly uses a certainty factor (CF) rule-based algorithm that can be calculated several symptoms to determine the inference solutions. The main issue for this TES included a prediction for some particular disease likelihood in the cases of new patients. CF is calculated based on symptoms related to clinical signs in patients’ diagnoses. For some reason, this TES probably won’t predict uncertain things, such as particular disease likelihood of some diseases. So, supervised learning, such as linear regression, can solve this problem. We tried to analyse the existing TES for thyroid disorders due to modelling the regression equation to predict the thyroid abnormality particular disease likelihood, based on the symptoms’ CF value and its confidence level. We used multiple linear regression (MLR) and multiple polynomial regression (MPR) to analyse the best regression model to solve the problem. The results show that the MPR model indicates the best regression model for predicting particular disease likelihood of thyroid abnormality, supported by R-squared 94.7%, R-squared adjusted 94.4%, F-value 265.925, and p-value < 0.05, which are higher than MLR model. Our study proposed a foundation for expert system development by focusing more on machine learning expert system (MLES) analysis approaches than TES.

Comparing Machine Learning Strategies for SoH Estimation of Lithium-Ion Batteries Using a Feature-Based Approach

Lithium-ion batteries play a vital role in many systems and applications, making them the most commonly used battery energy storage systems. Optimizing their usage requires accurate state-of-health (SoH) estimation, which provides insight into the performance level of the battery and improves the precision of other diagnostic measures, such as state of charge. In this paper, the classical machine learning (ML) strategies of multiple linear and polynomial regression, support vector regression (SVR), and random forest are compared for the task of battery SoH estimation. These ML strategies were selected because they represent a good compromise between light computational effort, applicability, and accuracy of results. The best results were produced using SVR, followed closely by multiple linear regression. This paper also discusses the feature selection process based on the partial charging time between different voltage intervals and shows the linear dependence of these features with capacity reduction. The feature selection, parameter tuning, and performance evaluation of all models were completed using a dataset from the Prognostics Center of Excellence at NASA, considering three batteries in the dataset.