Principal Component Regression Model Research Articles

Retrieve sea surface salinity using principal component regression model based on SMOS satellite data

The sea surface salinity (SSS) is a key parameter in monitoring ocean states. Observing SSS can promote the understanding of global water cycle. This paper provides a new approach for retrieving sea surface salinity from Soil Moisture and Ocean Salinity (SMOS) satellite data. Based on the principal component regression (PCR) model, SSS can also be retrieved from the brightness temperature data of SMOS L2 measurements and Auxiliary data. 26 pair matchup data is used in model validation for the South China Sea (in the area of 4°–25°N, 105°–125°E). The RMSE value of PCR model retrieved SSS reaches 0.37 psu (practical salinity units) and the RMSE of SMOS SSS1 is 1.65 psu when compared with in-situ SSS. The corresponding Argo daily salinity data during April to June 2013 is also used in our validation with RMSE value 0.46 psu compared to 1.82 psu for daily averaged SMOS L2 products. This indicates that the PCR model is valid and may provide us with a good approach for retrieving SSS from SMOS satellite data.

Bayesian model averaging and principal component regression forecasts in a data rich environment

This study revisits the accuracy of the point and density forecasts of monthly US inflation and output growth that are generated using principal components regression (PCR) and Bayesian model averaging (BMA). I run a forecasting horse race between 24 BMA specifications and two PCR alternatives in an out-of-sample, 10-year rolling event evaluation. The differences in mean-square forecast errors between BMA and PCR are mostly insignificant but predictable. PCR methods perform best for predicting deviations of output and inflation from their expected paths, whereas BMA methods perform best for predicting “tail” events. This dichotomy implies that risk-neutral policy-makers may prefer the classical PCR approach, while the BMA approach would belong in the toolkit of a prudential, risk-averse forecaster.

Soft sensing of product quality in the debutanizer column with principal component analysis and feed-forward artificial neural network

In this work, data-driven soft sensors are developed for the debutanizer column for online monitoring of butane content in the debutanizer column bottom product. The data set consists of data for seven process inputs and one process output. The total process data were equally divided into a training set and a validation set using the Kennard–Stone maximal intra distance criterion. The training set was used to develop multiple linear regression, principal component regression and back propagation neural network models for the debutanizer column. Performances of the developed models were assessed by simulation with the validation data set. Results show that the neural network model designed using Levenberg–Marquardt algorithm is capable of estimating the product quality with nearly 95% accuracy. The performance of the neural network model reported in this article is found to be better than the performances of least square support vector regression and standard support vector regression models reported in the literature earlier.

The Benefit of Including Rarely-Used Species in Dendroclimatic Reconstructions: A Case Study UsingJuglans nigrain South-Central Indiana, USA

ABSTRACT The benefit of using multiple species in dendroclimatic reconstructions in the eastern U.S. has been demonstrated. However, the benefit of including rarely-used species in multispecies reconstructions has been little explored. This paper shows the utility of using a rarely-used species in dendrochronology, Juglans nigra, in a multispecies Palmer Drought Severity Index (PDSI) reconstruction at a site in southern Indiana. First, the crossdating J. nigra is established, followed by determining the climate response. The standardized J. nigra chronology is then compared with co-occurring standardized species chronologies (Quercus alba, Quercus rubra, and Liriodendron tulipifera) reported in Maxwell et al. (2015). Using a principal component regression model, the bi-weights of each species were calculated to determine how much J. nigra contributed to the explanatory power of the model. J.nigra had a high interseries correlation (0.604) and mean sensitivity (0.304) and a strong correlation with summer P...

Correlation between bone marrow dose volumes and acute hematological toxicity in postoperative gynecological cancer patients.

Objective:To identify the association between radiation dose volume and acute hematological toxicity (HT) in postoperative gynecological cancer patients receiving whole pelvic radiotherapy (RT) or intensity-modulated RT (IMRT), a principal component regression model was used to calculate HT.Methods:Women (n=100) receiving with or without chemotherapy RT were retrospectively analyzed, 52 of whom received chemotherapy (paclitaxel and nedaplatin). The pelvis and lumbar vertebrae, defined as the prolong-pelvic bone marrow, were divided into the (1) combined ilium, ischium and pubis and the (2) lumbar vertebrae and the sacrum. The V5-V40 of subsides was calculated. The complete blood counts were recorded weekly. The principal component analysis was performed on volumes which generated the principal components (PCs), followed by using a logistic regression model.Results:Forty-seven patients presented with grade 2-3 HT during RT. Chemotherapy increased the incidence of HT compared with RT alone (70.21% vs. 29.79%; p=0.001). Fifty-three patients with persistent HT developed more serious HT at an earlier stage of RT. The chemotherapy cycles and three PCs associated with grade 2-3 HT was identified to form the resulting principal logistic regression model.Conclusion:A new method to calculate the NTCP was achieved by PCs logistic regression.

Spectroscopic-Based Chemometric Models for Quantifying Low Levels of Solid-State Transitions in Extended Release Theophylline Formulations

Variations in the solid state form of a pharmaceutical solid have profound impact on the product quality and clinical performance. Quantitative models that allow rapid and accurate determination of polymorphic changes in pharmaceutical products are essential in ensuring product quality throughout its lifecycle. This study reports the development and validation of chemometric models of Raman and near infrared spectroscopy (NIR) for quantifying the extent of pseudopolymorphic transitions of theophylline in extended release formulations. The chemometric models were developed using sample matrices consisting of the commonly used excipients and at the ratios in commercially available products. A combination of scatter removal (multiplicative signal correction and standard normal variate) and derivatization (Savitzky-Golay second derivative) algorithm were used for data pretreatment. Partial least squares and principal component regression models were developed and their performance assessed. Diagnostic statistics such as the root mean square error, correlation coefficient, bias and Q2 were used as parameters to test the model fit and performance. The models developed had a good fit and performance as shown by the values of the diagnostic statistics. The model diagnostic statistics were similar for MSC-SG and SNV-SG treated spectra. Similarly, PLSR and PCR models had comparable performance. Raman chemometric models were slightly better than their corresponding NIR model. The Raman and NIR chemometric models developed had good accuracy and precision as demonstrated by closeness of the predicted values for the independent observations to the actual TMO content hence the developed models can serve as useful tools in quantifying and controlling solid state transitions in extended release theophylline products.

Prediction of winter precipitation over northwest India using ocean heat fluxes

The winter precipitation (December–February) over northwest India (NWI) is highly variable in terms of time and space. The maximum precipitation occurs over the Himalaya region and decreases towards south of NWI. The winter precipitation is important for water resources and agriculture sectors over the region and for the economy of the country. It is an exigent task to the scientific community to provide a seasonal outlook for the regional scale precipitation. The oceanic heat fluxes are known to have a strong linkage with the ocean and atmosphere. Henceforth, in this study, we obtained the relationship of NWI winter precipitation with total downward ocean heat fluxes at the global ocean surface, 15 regions with significant correlations are identified from August to November at 90 % confidence level. These strong relations encourage developing an empirical model for predicting winter precipitation over NWI. The multiple linear regression (MLR) and principal component regression (PCR) models are developed and evaluated using leave-one-out cross-validation. The developed regression models are able to predict the winter precipitation patterns over NWI with significant (99 % confidence level) index of agreement and correlations. Moreover, these models capture the signals of extremes, but could not reach the peaks (excess and deficit) of the observations. PCR performs better than MLR for predicting winter precipitation over NWI. Therefore, the total downward ocean heat fluxes at surface from August to November are having a significant impact on seasonal winter precipitation over the NWI. It concludes that these interrelationships are more useful for the development of empirical models and feasible to predict the winter precipitation over NWI with sufficient lead-time (in advance) for various risk management sectors.

Remaining Useful Life Estimation with Lamb-Wave Sensors Based on Wiener Process and Principal Components Regression

This paper proposes a new approach for predicting the remaining useful life of a structure from Lamb wave sensor data using principal component regression and Wiener process degradation modeling. Principal component regression is used for extracting damage sensitive features of a Lamb wave sensor signal and establish a relation between the features and measured areas. A Wiener process is developed to model the random delamination growth and predict future damage and failure time probability. We use minimum squared prediction error to find the optimum number of features and the window inside a Lamb wave signal that is most sensitive to the fatigue damage. Method is illustrated on real sensor data collected under run-to-failure fatigue loading conditions.

Relating multi-sequence longitudinal intensity profiles and clinical covariates in incident multiple sclerosis lesions.

The formation of multiple sclerosis (MS) lesions is a complex process involving inflammation, tissue damage, and tissue repair — all of which are visible on structural magnetic resonance imaging (MRI) and potentially modifiable by pharmacological therapy. In this paper, we introduce two statistical models for relating voxel-level, longitudinal, multi-sequence structural MRI intensities within MS lesions to clinical information and therapeutic interventions: (1) a principal component analysis (PCA) and regression model and (2) function-on-scalar regression models. To do so, we first characterize the post-lesion incidence repair process on longitudinal, multi-sequence structural MRI from 34 MS patients as voxel-level intensity profiles. For the PCA regression model, we perform PCA on the intensity profiles to develop a voxel-level biomarker for identifying slow and persistent, long-term intensity changes within lesion tissue voxels. The proposed biomarker's ability to identify such effects is validated by two experienced clinicians (a neuroradiologist and a neurologist). On a scale of 1 to 4, with 4 being the highest quality, the neuroradiologist gave the score on the first PC a median quality rating of 4 (95% CI: [4,4]), and the neurologist gave the score a median rating of 3 (95% CI: [3,3]). We then relate the biomarker to the clinical information in a mixed model framework. Treatment with disease-modifying therapies (p < 0.01), steroids (p < 0.01), and being closer to the boundary of abnormal signal intensity (p < 0.01) are all associated with return of a voxel to an intensity value closer to that of normal-appearing tissue. The function-on-scalar regression model allows for assessment of the post-incidence time points at which the covariates are associated with the profiles. In the function-on-scalar regression, both age and distance to the boundary were found to have a statistically significant association with the lesion intensities at some time point. The two models presented in this article show promise for understanding the mechanisms of tissue damage in MS and for evaluating the impact of treatments for the disease in clinical trials.

Detection of Optimum Fermentation Time of Black CTC Tea Using a Voltammetric Electronic Tongue

A voltammetric electronic tongue (E-tongue) has been used to detect the optimum fermentation time (OFT) of crush tear curl black tea. An E-tongue with an array of five noble metal working electrodes has been developed for this purpose. A suitable large amplitude pulse voltammetric waveform is used and a simple feature extraction technique based on the charge activity of working electrodes is proposed. The principal component analysis reveals a significant correlation between the E-tongue responses and the biochemical changes during various stages of fermentation. Principal component regression models are developed to assess the effectiveness of the features derived from E-tongue response to model the temporal variations of theaflavins, a key quality compound of tea, produced during the fermentation process. The OFTs detected by the E-tongue correlate by a factor of 0.97 with the estimates obtained from an ultraviolet-visible spectrophotometer-based reference method.

3-D Fingertip Touch Force Prediction Using Fingernail Imaging With Automated Calibration

This paper presents an automated routine for calibrating a fingernail imaging system, with the intent of predicting fingerpad forces. The system uses a magnetic levitation haptic device to apply forces to the human fingerpad while recording images of the nail and surrounding skin. A novel force controller is implemented to interact stably with the human fingerpad. The data are used to calibrate a principal component regression model relating pixel intensity to 3-D force. Using data from this automated routine, this model simultaneously predicts 3-D force with an RMS error of 0.56 $\pm$ 0.03 N (7.7% of the full range of forces).

Optimizing Fingernail Imaging Calibration for 3D Force Magnitude Prediction.

This paper discusses the optimization of a fingernail imaging system for predicting fingerpad force. The effects of lighting coloration, calibration grid, and force prediction model on the registration process and force prediction accuracy of fingernail imaging are investigated. White and green LEDs are found to produce statistically similar effects on registration error and force prediction results across all three directions of force. Two calibration grids are implemented, with no statistically significant difference in either registration or force prediction between the Cartesian and cylindrical grid designs. Of the five force prediction models investigated, a principal component regression model based on the pixel intensity eigenvectors estimates the force with the greatest accuracy. This EigenNail Magnitude Model simultaneously estimates force in all three directions with RMS error with 95 percent confidence interval of 0.55 ± 0.02 N (7.6 percent of the full force range). These results indicate a set of optimal parameter choices for the calibration of a fingernail imaging system.

Process analytical technologies in food industry - challenges and benefits: A status report and recommendations.

Process analytical technologies in food industry - challenges and benefits: A status report and recommendations.

Modeling of ammonia emission in the USA and EU countries using an artificial neural network approach.

Ammonia emissions at the national level are frequently estimated by applying the emission inventory approach, which includes the use of emission factors, which are difficult and expensive to determine. Emission factors are therefore the subject of estimation, and as such they contribute to inherent uncertainties in the estimation of ammonia emissions. This paper presents an alternative approach for the prediction of ammonia emissions at the national level based on artificial neural networks and broadly available sustainability and economical/agricultural indicators as model inputs. The Multilayer Perceptron (MLP) architecture was optimized using a trial-and-error procedure, including the number of hidden neurons, activation function, and a back-propagation algorithm. Principal component analysis (PCA) was applied to reduce mutual correlation between the inputs. The obtained results demonstrate that the MLP model created using the PCA transformed inputs (PCA-MLP) provides a more accurate prediction than the MLP model based on the original inputs. In the validation stage, the MLP and PCA-MLP models were tested for ammonia emission predictions for up to 2 years and compared with a principal component regression model. Among the three models, the PCA-MLP demonstrated the best performance, providing predictions for the USA and the majority of EU countries with a relative error of less than 20%.

Application of NIR chemometric methods for quantification of the crystalline fraction of warfarin sodium in drug product

Monitoring of the physical state of warfarin sodium (WS) in products is essential for minimizing product quality variability in order to ensure consistent clinical performance. This study reports the development of chemometric models for quantifying the crystalline and amorphous fractions of WS in commercial drug products using NIR spectroscopy. Formulations based on commercially available products with different API to excipient ratio were used for the study. For each content, two formulations containing either lactose monohydrate or lactose anhydrous as the predominant formulation excipient were prepared. Two formulations containing either 100% amorphous WS (AWS) or crystalline WS (CWS) were prepared and mixed in various ratios to obtain sample matrices containing AWS/CWS 0–100%. The uniformity of the samples was confirmed by near infrared chemical imaging. Data were mathematically pretreated by multiplicative signal correction and Savitzky–Golay second derivative. Principal component regression and partial least square regression models were developed from mathematically treated data. All the models showed linear trend for amorphous and crystalline fractions of the WS as indicated by correlation and R2 > 0.99 and >0.98, respectively. The models demonstrated good performance parameters with a low-root mean squared error, standard error and bias. The model predicted CWS and AWS contents were in very close agreement with the actual values. The study indicated the utility of NIR chemometric methods in quantification of the crystalline and/or amorphous fraction of WS in its products.

A New Sector Rotation Strategy and its Performance Evaluation: Based on a Principal Component Regression Model

Sector rotation strategy is very important in the equity portfolio management. The paper builds a principal component regression model to design a sector rotation strategy, and uses a four-factor model to evaluate its performance in a frictional market. We concretely analyzes the emerging Shanghai security market, and finds that the principal component regression model could predict the future sector returns relatively precisely, and the sector rotation strategy could obtain a significant positive alpha return even with short sale constraints and transaction costs, and its performance mainly attributes to the market factor and sector momentum factor.

Principal component selection via adaptive regularization method and generalized information criterion

The principal component analysis has been widely used in various fields of research (e.g., bioinformatics, medical statistics, etc.), especially high dimensional data analysis. Although crucial components selection is a vital matter in principal components analysis, relatively little attention was paid to this issue. The existing studies for principal component analysis were based on ad-hoc methods (e.g., method with cumulative percent variance or average eigenvalue). We propose a novel method for selecting principal component based on L\(_{1}\)-type regularized regression modeling. In order to effectively perform for principal component regression, we consider adaptive L\(_{1}\)-type penalty based on singular values of components, and propose adaptive penalized principal component regression. The proposed method can perform feature selection incorporating explanation power of components for not only high-dimensional predictor variables but also response variable. In sparse regression modeling, choosing the regularization parameter is a crucial issue, since feature selection and estimation heavily depend on the selected regularization parameter. We derive a model selection criterion for choosing the regularization parameter of the proposed adaptive L\(_{1}\)-type regularization method in line with a generalized information criterion. Monte Carlo simulations and real data analysis demonstrate that the proposed modeling strategies outperform for principal component regression modeling.

Mixture Bayesian Regularization of PCR Model and Soft Sensing Application

In this paper, a Bayesian regularization mechanism is provided for automatically determining the number of latent variables in the probabilistic principal component regression (PPCR) model. Different from the unsupervised principal-component-analysis model, the response variable is incorporated for the supervision of selecting latent variables in the PPCR model. By introducing two hyperparameter vectors, the effectiveness of each latent variable can be well measured and controlled. For the mixture form of the PPCR model, a corresponding mixture Bayesian regularization strategy is further developed to control the dimensionality of latent variables. The expectation–maximization algorithm is employed for the parameter learning of both single and mixture Bayesian regularization models. Two probabilistic soft sensors are then developed for the online estimation of key variables in industrial processes, the performances of which are evaluated through two case studies. Compared to the single Bayesian regularization model, the mixture model shows stronger soft sensing abilities in nonlinear and multimode processes.

Discrimination and chemical characterization of different Paeonia lactifloras (Radix Paeoniae Alba and Radix Paeoniae Rubra) by infrared macro-fingerprint analysis-through-separation

Paeonia lactiflora, a commonly used herbal medicine (HM) in Traditional Chinese Medicine (TCM), mainly has two species, Radix Paeoniae Alba (RPA) and Radix Paeoniae Rubra (RPR), for different clinical applications in TCM. For expounding the chemical profile of RPA and RPR and ensuring the clinical efficacy and safety, an infrared macro-fingerprint analysis-through-separation method integrated with statistical pattern recognition was developed to analyze and discriminate the two Paeonia lactifloras. In IR spectra, the major difference between the two was in the range of 1200–900 cm−1: the strongest peak of RPA was at 1024 cm−1, while that of RPR was 1049 cm−1. The difference was magnified in second derivative spectra. The findings were further verified by investigating the separation process of total glucosides, stepwisely monitored by both of IR and UPLC-MS/MS. Simultaneously, the aqueous extracts of RPA and RPR had been separated continuously to acquire the comprehensively hierarchical chemical characteristics for undoubtedly identification and subsequently discrimination of the two herbs. Moreover, 60 batches of the two HMs (30 for each) were objectively classified by principal component regression (PCR) model based on IR macro-fingerprints.

Bayesian Typhoon Track Prediction Using Wind Vector Data

In this paper we predict the track of typhoons using a Bayesian principal component regression model based on wind field data. Data is obtained at each time point and we applied the Bayesian principal component regression model to conduct the track prediction based on the time point. Based on regression model, we applied to variable selection prior and two kinds of prior distribution; normal and Laplace distribution. We show prediction results based on Bayesian Model Averaging (BMA) estimator and Median Probability Model (MPM) estimator. We analysis 8 typhoons in 2006 using data obtained from previous 6 years (2000–2005). We compare our prediction results with a moving-nest typhoon model (MTM) proposed by the Korea Meteorological Administration. We posit that is possible to predict the track of a typhoon accurately using only a statistical model and without a dynamical model.