Kernel Ridge Regression Model Research Articles

Heterogeneous Transfer Learning of Electrohydrodynamic Printing Under Zero-Gravity Toward In-Space Manufacturing

Abstract As we continue to commercialize space and mature in-space manufacturing (ISM) processes, there is a strong need to transfer the knowledge we learn from experiments on the ground to zero-gravity environments. Physics-motivated manufacturing processes, like additive manufacturing, experience a shift in fabrication parameters due to the absence of gravity and the change of environments. Thus, we found traditional machine learning methods are not capable of addressing this domain shift and present a transfer learning scheme as a solution in this paper. We tested a kernel ridge regression model built for heterogeneous transfer learning (KRR-HeITL) on data from the electrohydrodynamic inkjet printing (EHD printing) process. EHD printing is a process that uses electrical force to control material flows, thus achieving the fabrication of electronics without requiring gravity. Our team has successfully conducted three rounds of parabolic flights to validate this technology for ISM. We trained on multiple datasets built from on-ground experiments and tested using zero-gravity printing data obtained from parabolic flight tests. Measurements of the Taylor cone both on-ground and in zero-gravity were taken and exploited as a part of the training data. We found that our method obtains good interpolation accuracy (MAPE 3.85%) compared to traditional machine learning methods (MAPE 16.84%) for predicting the printed line width. We concluded that the KRR-HeITL method is well suited for zero-gravity domain shifts of EHD printing parameters. This study paves the way for future predictions of ISM parameters when there are only on-ground experiments or very limited zero-gravity datasets for a given process.

Predicting wear resistance of high-carbon Cr-V alloy steel based on machine learning

The microstructure of Cr-V wear-resistant alloy steel was characterized by its complexity, with the evaluation of its wear resistance exhibiting multidimensional, strongly coupled, and nonlinear attributes. Contrary to the conventional research paradigm that relied on microstructure analysis and performance comparison, accurate prediction of wear resistance based on data-driven approaches was found to significantly expedite the research and development process for materials. In this study, microstructural analysis demonstrated that the grid-like arrangement of M7C3, MC, and M2C eutectic carbides, along with the dispersed granular M23C6 carbides within the matrix, significantly enhanced material hardness. The uniform distribution of carbides played a critical role in resisting abrasive wear. This investigation further introduced the mass score ratios of (Cr+V)/C and V/Cr as evaluation indices for carbides to optimize the characteristic dataset. The findings unequivocally revealed that optimization of the feature set significantly enhanced the performance of the predictive models, with the correlation coefficient (R) value showing an increase of 3.17–25.18 %, while the root mean square error (RMSE) value decreased by 33.26–71.15 %. Compared to the Random Forest (RF), Gradient Boosting Trees (GBT), and Kernel Ridge Regression (KRR) models，the Support Vector Regression(SVR) model demonstrated preeminent performance, evidenced by an R value of 0.979 and an RMSE value of 0.866 mg. The application of the trained SVR model to predict new samples yielded satisfactory results, with the predicted wear resistance rankings being entirely consistent with the experimental outcomes and exhibiting high prediction accuracy. The model possessed robust quantitative predictive capabilities, enabling rapid evaluation of the wear resistance of this category of steel.

Detecting bioactive compound contents in Dancong tea using VNIR-SWIR hyperspectral imaging and KRR model with a refined feature wavelength method

Hyperspectral imaging (HSI) provides opportunity for non-destructively detecting bioactive compounds contents of tea leaves and high detection accuracy require extracting effective features from the complex hyperspectral data. In this paper, we proposed a feature wavelength refinement method called interval band selecting-competitive adaptive reweighted sampling-fusing (IBS-CARS-Fusing) to extract feature wavelengths from visible-near-infrared (VNIR) and short-wave-near-infrared (SWIR) hyperspectral images. Combined with the proposed IBS-CARS-Fusing method, a kernel ridge regression (KRR) model was established to predict the contents of bioactive compounds including chlorophyll a, chlorophyll b, carotenoids, tea polyphenols, and amino acids in Dancong tea. It was revealed that the IBS-CARS-Fusing method can improve Rp2 of KRR model for these bioactive compounds by 4.77%, 4.60%, 6.74%, 15.52%, and 13.10%, respectively, and Rp2 of the model reached high values of 0.9500, 0.9481, 0.8946, 0.8882, and 0.8622. Additionally, a leaf compound mass per area thermal map was used to visualize the spatial distribution of the compounds.

Advanced modeling of pharmaceutical solubility in solvents using artificial intelligence techniques: assessment of drug candidate for nanonization processing.

This research is an analysis of multiple regression models developed for predicting ketoprofen solubility in supercritical carbon dioxide under different levels of T(K) and P(bar) as input features. Solubility of the drug was correlated to pressure and temperature as major operational variables. Selected models for this study are Piecewise Polynomial Regression (PPR), Kernel Ridge Regression (KRR), and Tweedie Regression (TDR). In order to improve the performance of the models, hyperparameter tuning is executed utilizing the Water Cycle Algorithm (WCA). Among, the PPR model obtained the best performance, with an R2 score of 0.97111, alongside an MSE of 1.6867E-09 and an MAE of 3.01040E-05. Following closely, the KRR model demonstrated a good performance with an R2 score of 0.95044, an MSE of 2.5499E-09, and an MAE of 3.49707E-05. In contrast, the TDR model produces a lower R2 score of 0.84413 together with an MSE of 7.4249E-09 and an MAE of 5.69159E-05.

Optical properties of NiO films: Effect of nitrogen-doping, substrate temperature and band gap estimation using machine learning

In this study, nickel oxide (NiO) thin films were synthesized on glass substrates using RF magnetron sputtering with varying nitrogen (N) doping ratio and substrate temperatures to explore modifications in their structural, morphological, and optical properties. The films were prepared using a high-purity NiO target under controlled sputtering conditions. Structural analysis by X-ray diffraction revealed an improvement in crystallinity in the (200) direction with increasing N ratio. In contrast, higher N ratio led to the suppression of (111) and (220) peaks, indicating a significant influence of N on the crystal structure and orientation. The films’ thickness and morphology, examined using scanning electron microscopy and energy-dispersive X-ray spectroscopy, showed uniform and homogeneous growth with smooth surface topologies. Optical properties, assessed by UV–Vis-NIR spectrophotometry, demonstrated a decrease in transmittance and a redshift in the absorption edge with increased N doping, corresponding to a narrowing of the energy bandgap from 3.7 eV to 3.45 eV. This bandgap reduction is attributed to N incorporation substituting oxygen sites, introducing defect states within the band structure. Additionally, the impact of substrate temperature on film growth enhanced crystallinity and orientation along the (111) plane at higher temperatures, with a simultaneous reduction in film thickness due to increased adatom mobility and potential thermal decomposition. The evaluation of Kernel Ridge Regression (KRR) and Ridge Regression (RR) models revealed their effectiveness in predicting band gap values for thin films at varying substrate temperatures and thicknesses. While RR excelled in predicting a band gap of 3.6 eV for a film with a substrate temperature of 24 °C and a thickness of 112.7 nm, KRR outperformed in predicting a band gap of 3.65 eV for a film with a substrate temperature of 24 °C and a thickness of 107 nm. These findings elucidate the dual influence of N doping and substrate temperature on enhancing the functional properties of NiO thin films, promising for applications in optoelectronic devices and gas sensors.

Using Data Science Tools to Reveal and Understand Subtle Relationships of Inhibitor Structure in Frontal Ring-Opening Metathesis Polymerization.

The rate of frontal ring-opening metathesis polymerization (FROMP) using the Grubbs generation II catalyst is impacted by both the concentration and choice of monomers and inhibitors, usually organophosphorus derivatives. Herein we report a data-science-driven workflow to evaluate how these factors impact both the rate of FROMP and how long the formulation of the mixture is stable (pot life). Using this workflow, we built a classification model using a single-node decision tree to determine how a simple phosphine structural descriptor (Vbur-near) can bin long versus short pot life. Additionally, we applied a nonlinear kernel ridge regression model to predict how the inhibitor and selection/concentration of comonomers impact the FROMP rate. The analysis provides selection criteria for material network structures that span from highly cross-linked thermosets to non-cross-linked thermoplastics as well as degradable and nondegradable materials.

Cross-scale analysis for the thermo-hydro-mechanical (THM) effects on the mechanical behaviors of fractured rock: Integrating mesostructure-based DEM modeling and machine learning

The excavation process induces the emergence of damage zones in the on-site engineering rock masses, significantly affecting the parameter evaluation of surrounding rock masses. Therefore, an in-depth understanding of the thermo-hydro-mechanical (THM) effects on the cross-scale failure process of fractured rock becomes progressively crucial in natural energy exploitation projects. In this study, a coupled thermo-hydro-mesostructure-based DEM (T-H-MSBM) model was developed to reconstruct rock microstructures and distinguish the THM responses of varying mineral grains, micro-defects and cracks in fractured granite. Five sets of numerical fractured granite in terms of different damage degrees were generated by restoring the form of excavation damage on site, and a series of compression simulations were conducted on the T-H-MSBMs under coupled temperature (25–250 °C), pore pressure (0–15 MPa) and confining pressure (15 MPa) conditions. Based on the cross-scale failure analysis on the fractured granite during the THM-compression loading process, the interplay between THM treatment and damage degree on the mechanical properties of fractured granite was revealed, and the main mechanisms affecting the varied macro mechanical properties were further discussed. Results indicate that both temperature and pore pressure exert the amplified deteriorating effect on the macro mechanical properties of fractured granite with increasing damage degree, while the temperature dependence becomes significantly more pronounced in the fractured granite with low damage degree. With an increase in the damage degree of fractured rock, the abundance of intra-mineral tensile cracks accelerates the initiation of pore pressure-induced cracks, and the coupled increase in pore pressure and temperature enhances the role of quartz-feldspar cracks in promoting the cracks in quartz. Finally, utilizing the optimized kernel ridge regression (KRR) model employed the grid search method, the correlation information database of rock macro-mechanical strength and THM factors were generated. The entire training time was less than 3 s, with the relative average errors of 0.7 MPa, which demonstrated the robust capability of machine learning method in predicting the coupled THM phenomena spanning multiple scales in geological materials and systems.

Development of a mathematical model for investigation of hollow-fiber membrane contactor for membrane distillation desalination

This research investigates the predictive modeling of a dataset containing parameters denoted by r(m), z(m), and T(K) which is temperature. The considered process is a membrane distillation (MD) for separation of compounds based on temperature gradient. A membrane contactor is used for the process, and the computations are performed in the context of computational fluid dynamics (CFD) and machine learning. The dataset, which is generated by CFD modeling and encompasses over 5,000 data points, is analyzed using three distinct regression models: Support Vector Machine (SVM), Deep Neural Network (DNN), and Kernel Ridge Regression (KRR). Hyperparameter tuning is performed employing the Stochastic Fractal Search (SFS) algorithm. Our findings unraveled the nuanced intricacies of each model's performance, gauged through a comprehensive set of metrics. The RMSE, MAPE, and R2 score collectively offer a robust evaluation framework. The Deep Neural Network (DNN) exhibits a compelling RMSE of 7.7001E-01, a remarkably low MAPE of 2.05131E-03, and an impressive R2 score of 0.97054. Meanwhile, the Support Vector Machine (SVM) showcases a notable RMSE of 1.7215E-01, a minimal MAPE of 2.90820E-04, and a remarkably high R2 score of 0.99839. On the other hand, the Kernel Ridge Regression (KRR) model presents an RMSE of 1.3588E + 00, a MAPE of 2.63550E-03, and an R2 score of 0.90042.

Comparison of Ridge and Kernel Ridge Models in Predicting Thermal Stability of Zn-MOF Catalysts

This study investigates machine learning-based quantitative structure-property relationship (QSPR) models for predicting the thermal stability of zinc metal-organic frameworks (Zn-MOF). Utilizing a dataset comprising 151 Zn-MOF compounds with relevant molecular descriptors, ridge (R) and kernel ridge (KR) regression models were developed and evaluated. The results demonstrate that the R model outperforms the KR model in terms of prediction accuracy, with the R model exhibiting exceptional performance (R² = 0.999, RMSE = 0.0022). While achieving high accuracy, opportunities for further improvement exist through hyperparameter optimization and exploration of polynomial functions. This research underscores the potential of ML-based QSPR models in predicting the thermal stability of Zn-MOF compounds and highlights avenues for future investigation to enhance model accuracy and applicability in materials science.

Daily scale air quality index forecasting using bidirectional recurrent neural networks: Case study of Delhi, India

This research was established to accurately forecast daily scale air quality index (AQI) which is an essential environmental index for decision-making. Researchers have projected different types of models and methodologies for AQI forecasting, such as statistical techniques, machine learning (ML), and most recently deep learning (DL) models. The modelling development was adopted for Delhi city, India which is a major city with air pollution issues simialir to entire urban cities of India especially during winter seasons. This research was predicted AQI using different versions of DL models including Long-Short Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM) and Bidirectional Recurrent Neural Networks (Bi-RNN) in addition to Kernel Ridge Regression (KRR). Results indicated that Bi-RNN model consistently outperformed the other models in both training and testing phases, while the KRR model consistently displayed the weakest performance. The outstanding performance of the models development displayed the requirement of adequate data to train the models. The outcomes of the models showed that LSTM, BI-LSTM, KRR had lower performance compared with Bi-RNN models. Statistically, Bi-RNN model attained maximum cofficient of determination (R2 = 0.954) and minimum root mean square error (RMSE = 25.755). The proposed model in this research revealed the robust predictable to provide a valuable base for decision-making in the expansion of combined air pollution anticipation and control policies targeted at addressing composite air pollution problems in the Delhi city.

A hybrid machine-learning model for solar irradiance forecasting

Abstract Nowcasting and forecasting solar irradiance are vital for the optimal prediction of grid-connected solar photovoltaic (PV) power plants. These plants face operational challenges and scheduling dispatch difficulties due to the fluctuating nature of their power output. As the generation capacity within the electric grid increases, accurately predicting this output becomes increasingly essential, especially given the random and non-linear characteristics of solar irradiance under variable weather conditions. This study presents a novel prediction method for solar irradiance, which is directly in correlation with PV power output, targeting both short-term and medium-term forecast horizons. Our proposed hybrid framework employs a fast trainable statistical learning technique based on the truncated-regularized kernel ridge regression model. The proposed method excels in forecasting solar irradiance, especially during highly intermittent weather periods. A key strength of our model is the incorporation of multiple historical weather parameters as inputs to generate accurate predictions of future solar irradiance values in its scalable framework. We evaluated the performance of our model using data sets from both cloudy and sunny days in Seattle and Medford, USA and compared it against three forecasting models: persistence, modified 24-hour persistence and least squares. Based on three widely accepted statistical performance metrics (root mean squared error, mean absolute error and coefficient of determination), our hybrid model demonstrated superior predictive accuracy in varying weather conditions and forecast horizons.

Forest yield prediction under different climate change scenarios using data intelligent models in Pakistan.

This study aimed to develop and evaluate data driven models for prediction of forest yield under different climate change scenarios in the Gallies forest division of district Abbottabad, Pakistan. The Random Forest (RF) and Kernel Ridge Regression (KRR) models were developed and evaluated using yield data of two species (Blue pine and Silver fir) as an objective variable and climate data (temperature, humidity, rainfall and wind speed) as predictive variables. Prediction accuracy of both the models were assessed by means of root mean squared error (RMSE), mean absolute error (MAE), correlation coefficient (r), relative root mean squared error (RRMSE), Legates-McCabe's (LM), Willmott's index (WI) and Nash-Sutcliffe (NSE) metrics. Overall, the RF model outperformed the KRR model due to its higher accuracy in forecasting of forest yield. The study strongly recommends that RF model should be applied in other regions of the country for prediction of forest growth and yield, which may help in the management and future planning of forest productivity in Pakistan.

MedOptNet: Meta-Learning Framework for Few-Shot Medical Image Classification.

In the medical research domain, limited data and high annotation costs have made efficient classification under few-shot conditions a popular research area. This paper proposes a meta-learning framework, termed MedOptNet, for few-shot medical image classification. The framework enables the use of various high-performance convex optimization models as classifiers, such as multi-class kernel support vector machines, ridge regression, and other models. End-to-end training is then implemented using dual problems and differentiation in the paper. Additionally, various regularization techniques are employed to enhance the model's generalization capabilities. Experiments on the BreakHis, ISIC2018, and Pap smear medical few-shot datasets demonstrate that the MedOptNet framework outperforms benchmark models. Moreover, the model training time is also compared to prove its effectiveness in the paper, and an ablation study is conducted to validate the effectiveness of each module.

A new extraction and grading method for underwater topographic photons of photon-counting LiDAR with different observation conditions

ABSTRACT Spaceborne photon-counting light detection and ranging (LiDAR) have been extensively applied in shallow-water bathymetry. The density of underwater topographic photons (UTP) varies and is discontinuous due to sunlight noise, beam intensity, and seabed reflectivity, which differ from the land photon distribution due to the attenuation of water. Therefore, a general method for extracting and grading UTP is still lacking. We propose an active contour method combined with a variable convolution kernel method to calculate the photon range by considering the energy contributions of adjacent photons. Adaptive parameters under different observation conditions were determined to obtain the optimal convolution kernel using a kernel ridge regression model. This implies that the number of photons contained in the buffer zone was largest after the extracted UTP was fitted to a curve. Quantitative and qualitative verifications proved that the method performed well under different conditions. The photons obtained by the energy functional and the curve obtained by the fitting method were then used to grade the photons. Finally, an online developed UTP dataset and extraction framework were proposed to provide an applicable method for current and subsequent spaceborne photon-counting LiDAR.

Additive manufacturing of hybrid piezoelectric/magnetic self-sensing actuator using pellet extrusion and immersion precipitation with statistical modelling optimization

Additive manufacturing is a growing field of fast reliable processing as it can fabricate complex designs, both internally and externally. Multi-stimuli-responsive/-functional polymers can respond to numerous stimuli and execute multiple tasks. A thin flexible hybrid piezoelectric–magnetic self-sensing actuator (HPMSA) is printed utilizing pellet extrusion and immersion precipitation 3D printing (ip3DP) to enhance its performance as both a sensor and an actuator. Utilizing molecular interactions and pores created within the structure, ip3DP showcased a more stable and effective self-sensing actuator than pellet extruded samples. Additionally, HPMSA fabricated using either additive manufacturing methods had a higher overall crystal content of 62.1 % compared to the conventional process of compression molding and mechanical stretching, highlighting its scale-up fabrication whilst promoting piezoelectric crystals. For optimization, kernel ridge regression model was utilized to predict the optimal ip3DP condition, which was experimentally validated. As a vibration damper, the ip3DP HPMSA with an optimized geometry showcased an effective high voltage sensing output of 13 mV/g and maximum weighted damping of 1.8 m/s2, lowering passenger health risks to “caution zone” in high vibration environments. The thin and flexible HPMSA provides understanding into multi-stimuli/-functional materials, simultaneous alignment, and vibration control.

Machine learning based prediction model for the pile bearing capacity of saline soils in cold regions

The difficulty in determining the bearing capacity of pile foundations in saline soil environments in cold regions can pose a challenge when developing a bearing capacity prediction model. To address this, the study uses data from the construction of the De Xiang Expressway project in Qinghai Province, China, and considers pile length, pile diameter, corrosion depth, and spalling thickness as influential parameters. We combine Python with ABAQUS to build a finite element model of a single pile in a salt marsh environment with random parameters and develop a database for machine learning. We use the kernel ridge regression (KRR) and multilayer perceptron (MLP) algorithms to establish load-bearing capacity prediction models for vertical and horizontal loads, respectively. The SHAP (SHapley Additive exPlanations) method is employed to perform explanatory analysis on the prediction models, providing insight into influential parameter effects on prediction accuracy. The results demonstrate that the database established through the random parameter finite element method is suitable for machine learning. While the KRR model exhibits good performance in predicting vertical load-bearing capacity, the MLP model performs well in predicting horizontal load-bearing capacity. Moreover, the SHAP method effectively explains the importance of influential parameters on the prediction results and enhances the reliability of the prediction models. The vertical bearing capacity of a saline soil pile foundation is more affected by the depth of salt erosion than the thickness of spalling, while the horizontal bearing capacity is more affected by the thickness of spalling than the depth of salt erosion.

Prediction of transportation index for urban patterns in small and medium-sized Indian cities using hybrid RidgeGAN model

The rapid urbanization trend in most developing countries including India is creating a plethora of civic concerns such as loss of green space, degradation of environmental health, scarcity of clean water, rise in air pollution, and exacerbated traffic congestion resulting in significant delays in vehicular transportation. To address the intricate nature of transportation issues, many researchers and planners have analyzed the complexities of urban and regional road systems using transportation models by employing transportation indices such as road length, network density, accessibility, and connectivity metrics. This study addresses the complexities of predicting road network density for small and medium-sized Indian cities that come under the Integrated Development of Small and Medium Towns (IDSMT) project at a national level. A hybrid framework based on Kernel Ridge Regression (KRR) and the CityGAN model is introduced to predict network density using spatial indicators of human settlements. The major goal of this study is to generate hyper-realistic urban patterns of small and medium-sized Indian cities using an unsupervised CityGAN model and to study the causal relationship between human settlement indices (HSIs) and transportation index (network density) using supervised KRR for the real cities. The synthetic urban universes mimic Indian urban patterns and evaluating their landscape structures through the settlement indices can aid in comprehending urban landscape, thereby enhancing sustainable urban planning. We analyzed 503 real cities to find the actual relationship between the urban settlements and their road density. The nonlinear KRR model may help urban planners in deriving the network density for GAN-generated futuristic urban patterns through the settlement indicators. The proposed hybrid process, termed as RidgeGAN model, can gauge the sustainability of urban sprawl tied to infrastructure and transportation systems in sprawling cities. Analysis results clearly demonstrate the utility of RidgeGAN in predicting network density for different kinds of human settlements, particularly for small and medium Indian cities. By predicting future urban patterns, this study can help in the creation of more livable and sustainable areas, particularly by improving transportation infrastructure in developing cities.

Machine learning approach to predicting the macro-mechanical properties of rock from the meso-mechanical parameters

The determination of fundamental rock mechanical properties, uniaxial compression strength (UCS) and elastic modulus (E), constitutes a pivotal facet of rock engineering design. However, deriving these properties directly from standard laboratory tests on rock core samples can be challenging, especially when dealing with deep high-stress rock formations and weak fractured strata. Thus, it is crucial to establish a cost-effective and practical approach for predicting the macro-mechanical properties of rocks in situ. In this study, a machine learning approach was proposed to predict UCS and E by upscaling meso-mechanical parameters at particle scale in low-porosity crystalline rocks. To expand the correlation database of rock meso-macro mechanical properties, the meso-mechanical parameters, including the fracture toughness, tensile strength of the rock crystal interface, and the elastic modulus of rock crystal, were accurately measured, using a newly designed mechanical apparatus and a nanoindentation device. The grain-based models implemented in the combined finite discrete element method (FDEM-GBM) were developed based on these experimental results, and their reliability was validated though standard tests. Subsequently, a database, including 225 groups of data, was established using the numerical method. Five machine learning algorithms were applied to develop prediction models for UCS and E through data training in the database. Excellent performance improvement was achieved through the application of the grid-search method. The results indicate the optimized kernel ridge regression (KRR) and gaussian process regression (GPR) models demonstrated excellent performance with relative average errors of 4.9% and 1.1% in predicting UCS and E, respectively. Finally, the predicted values of UCS and E were compared with the experimental results, validating the feasibility of the optimized KRR and GPR models.

Machine learning estimation of reaction energy barriers

We describe a machine learning approach to approximate reaction energy barriers (E), requiring as input only estimates of geometry and energies of reactants and products. Using the dataset of Grambow, Pattanaik, and Green [Sci. Data 7 (137) (2020)] for reactions involving seven or fewer non-hydrogen atoms, 300 reaction features are computed, and an estimate of E is obtained by fitting a Kernel Ridge Regression (KRR) model with Laplacian kernel to a subset of Density Functional Theory reaction barriers. Our main interest is small energy barriers with the goal of modeling reactions in the interstellar medium and circumstellar envelope. We omitted reactions with E > 40 kcal mol−1 to obtain a subset of 5,276 reactions for 5-fold cross-validation. For this set, the KRR model predicts E with a mean absolute error of 4.13 kcal mol−1 and a root-mean square error of 6.02 kcal mol−1.

Predicting corrosion inhibition efficiencies of small organic molecules using data-driven techniques

Selecting effective corrosion inhibitors from the vast chemical space is not a trivial task, as it is essentially infinite. Fortunately, machine learning techniques have shown great potential in generating shortlists of inhibitor candidates prior to large-scale experimental testing. In this work, we used the corrosion responses of 58 small organic molecules on the magnesium alloy AZ91 and utilized molecular descriptors derived from their geometry and density functional theory calculations to encode their molecular information. Statistical methods were applied to select the most relevant features to the target property for support vector regression and kernel ridge regression models, respectively, to predict the behavior of untested compounds. The performance of the two supervised learning approaches were compared and the robustness of the data-driven models were assessed by experimental blind testing.