Hierarchical Bayesian Research Articles

Hierarchical Bayesian spectral regression with shape constraints for multi-group data

We propose a hierarchical Bayesian (HB) model for multi-group analysis with group–specific, flexible regression functions. The lower–level (within group) and upper–level (between groups) regression functions have hierarchical Gaussian process priors. HB smoothing priors are developed for the spectral coefficients. The HB priors smooth the estimated functions within and between groups. The HB model is particularly useful when data within groups are sparse because it shares information across groups, and provides more accurate estimates than fitting separate nonparametric models to each group. The proposed model also allows shape constraints, such as monotone, U and S–shaped, and multi-modal constraints. When appropriate, shape constraints improve estimation by recognizing violations of the shape constraints as noise. The model is illustrated by two examples: monotone growth curves for children, and happiness as a convex, U-shaped function of age in multiple countries. Various basis functions could also be used, and the paper also implements versions with B-splines and orthogonal polynomials.

Error modeling and hierarchical Bayesian fusion for spaceborne and ground radar rainfall data

The optimal fusion of multisource rainfall data can improve the accuracy and resolution of a posteriori estimates, which is beneficial for hydrological and meteorological applications. In this study, the errors of ground-based dual-polarization radar (GR) and dual-frequency precipitation radar (DPR) on global precipitation measurement (GPM) satellite rainfall estimation were modeled, and a hierarchical Bayesian (HB) framework was developed for GR and DPR rainfall data fusion. According to the matched optimal estimated GR rainfall and rain gauge data, the GR rainfall error was analyzed, and the error variance varying with rain type and intensity was described and modeled. The rain types were classified based on disdrometer and dual-polarization radar data. By comparing the DPR rainfall product with GR rainfall, the DPR rainfall error was divided into systematic and random errors. The systematic error was linearly fitted, and the probability distributions for the random error and the variable error parameters were analyzed and modeled. Because the parameters for GR and DPR rainfall errors varied with rain type and intensity, we developed a HB fusion approach considering the varying parameters in rainfall error modeling. The performance of the fusion algorithm was evaluated using matched GR and DPR cases, which showed that the HB fusion can effectively improve the consistency between the original DPR rainfall product and rain gauge data, reducing the relative bias by an average of 30% and improving the correlation by an average of 20%. Compared with the conventional Bayesian (CB) fusion algorithm, which is based on overall error modeling, the HB fusion results can improve the root mean square error by an average of 24% and relative bias by 14%.

EEG based emotion recognition by hierarchical bayesian spectral regression framework

Spectral regression (SR), a graph-based learning regression model, can be used to extract features from graphs to realize efficient dimensionality reduction. However, due to the SR method remains a regularized least squares problem and being defined in L2-norm space, the effect of artifacts in EEG signals cannot be efficiently resisted. In this work, to further improve the robustness of the graph-based regression models, we propose to utilize the prior distribution estimation in the Bayesian framework and develop a robust hierarchical Bayesian spectral regression framework (named HB-SR), which is designed with the hierarchical Bayesian ensemble strategies. In the proposed HB-SR, the impact of noises can be effectively reduced by the adaptive adjustment approach in model parameters with the data-driven manner. Specifically, in the current work, three different distributions have been elaborately designed to enhance the universality of the proposed HB-SR, i.e., Gaussian distribution, Laplace distribution, and Student-t distribution. To objectively evaluate the performance of the HB-SR framework, we conducted both simulation studies and emotion recognition experiments based on emotional EEG signals. Experimental results have consistently indicated that compared with other existing spectral regression methods, the proposed HB-SR can effectively suppress the influence of noises and achieve robust EEG emotion recognition.

Understanding Patient and Physician Perceptions Regarding Treatment-Related Adverse Events in First-Line Chronic Lymphocytic Leukemia (CLL)

Background Chronic Lymphocytic Leukemia (CLL) patients have various treatment options which are all associated with different adverse events. A limited number of studies have explored how patients and physicians perceive these adverse events in the frontline treatment setting. Understanding the similarities and differences in perceptions between physicians and CLL patients is an important factor in supporting shared decision-making. Aims The objective of this multi-country study was to understand the differences in patients' and physicians' perception on treatment-related adverse events in CLL. Methods CLL patients and physicians managing CLL (oncologists, hematologist-oncologists, and hematologists) were invited to complete an online survey. Participants were recruited from the US, UK, Germany, France, and Australia. The survey included a best-worst scaling (BWS) exercise to assess how ‘bothersome’ 14 potential adverse events associated with 1L CLL treatments were perceived. In a series of 18 BWS choice tasks, participants chose which was most ‘bothersome’ and which was ‘least bothersome’ among subsets of 4 adverse events. A hierarchical Bayesian (HB) model was used to estimate BWS scores for each treatment-related adverse event: BWS scores sum to 100 for each participant and a higher score indicates that an adverse event is relatively more ‘bothersome’. Results The study included 192 patients and 259 physicians. The top four most bothersome adverse events were the same for both patients and physicians, however, these were ranked in a different order (Table 1). Physicians ranked the ‘risk of cardiac events/atrial fibrillation/flutter that may require medical treatment and/or hospitalization’ and ‘risk of bleeding (hemorrhage) that requires medical intervention and/or hospitalization’ as their 2 most bothersome adverse events while patients ranked ‘Risk of infection, such as pneumonia or blood infections (sepsis) requiring medical intervention and/or hospitalization’ and ‘Risk of tumor lysis syndrome (TLS) which requires hospitalization’ as their 2 most bothersome adverse events (Table 1). The 2 least bothersome adverse events for both physicians and patients were ‘Risk of mild-to-moderate muscle, joint or bone pain (mild-to-moderate pain not interfering with daily activities) that might go away with medication’ and ‘Risk of mild-to-moderate headache (not limiting daily activities or limiting only instrumental activities such as cooking, cleaning, transportation, laundry, etc.) that might go away with medication’ (Table 1). Among patients who had experienced an adverse event (N=111), 62 (56%) reported that it had impacted their quality of life. These patients rated the following adverse events as more bothersome compared to those who had not experienced adverse events impacting quality of life: ‘risk of infection, such as pneumonia or blood infections requiring medical intervention and/or hospitalization’ (BWS scores=13.88 vs. 11.79) and ‘a side effect resulting in stopping the medication’ (9.46 vs. 6.00). These patients also rated the following adverse events as least bothersome: ‘risk of severe headache that might lead to hospitalization’ (9.55 vs. 11.68), ‘risk of severe vomiting that might lead to hospitalization’ (8.97 vs. 10.70), and ‘risk of severe muscle joint or bone pain’ (7.32 vs. 10.59), compared to those not experiencing adverse events impacting quality of life. Conclusions Patients' and physicians' perceptions were generally aligned on what adverse events related to 1L CLL treatments are most and least bothersome. However, the rank-ordering of the most bothersome adverse events varied between patients and physicians: patients were more concerned about a risk of infection or tumor lysis syndrome whereas physicians thought that a risk of cardiac event and the risk of bleeding were the most bothersome adverse events. Patients expressed hardly any concern about a risk of mild-to-moderate headache that could be resolved with medication. Interestingly, patients who had experienced a prior adverse event had differing results to patients who had not. These results should further inform and facilitate conversations between physicians and patients and improve shared decision-making when selecting a 1L regimen to treat CLL.

Exploring patient treatment decision making for osteoarthritis in the UAE: a cross-sectional adaptive choice-based conjoint study

ObjectiveTo assess osteoarthritis (OA) patients’ preferences for pharmaceutical treatment via Adaptive Choice-Based Conjoint (ACBC) method.MethodsA United Arab Emirates (UAE) based Patient and Public Involvement (PPI) group designed the ACBC questionnaire with 10 attributes and 34 levels. The questionnaire was developed using Sawtooth Software and analyzed through Hierarchical Bayesian (HB). Results were standardized using Z-score via SPSS.ResultsStudy participants were 1030 OA patients, 83.6% aged 50 or older and 83.4% female. The avoidance of medication’s side effects accounted for 66% relative importance compared to 6% relative importance for the medication’s benefits. The “way of taking the medicine” attribute had the highest coefficient of variation (70%) and the four side effect attributes “risk of gastric ulcer, addiction, kidney and liver impairment, and heart attacks and strokes” had a coefficient of variation from 18 to 21%.ConclusionsArab OA patients are similar to other ethnic groups in trading-off benefits and side effects and consistently prioritizing the avoidance of medications’ side effects. Although the “Way of taking medicine” was the least important attribute it was associated with the highest variation amongst patients. OA patients also prefer prescribed medications to internet-purchased and over-the-counter options.

Long-term forecasting of a motor outcome following rehabilitation in chronic stroke via a hierarchical bayesian dynamic model

BackgroundGiven the heterogeneity of stroke, it is important to determine the best course of motor therapy for each patient, i.e., to personalize rehabilitation based on predictions of long-term outcomes. Here, we propose a hierarchical Bayesian dynamic (i.e., state-space) model (HBDM) to forecast long-term changes in a motor outcome due to rehabilitation in the chronic phase post-stroke.MethodsThe model incorporates the effects of clinician-supervised training, self-training, and forgetting. In addition, to improve forecasting early in rehabilitation, when data are sparse or unavailable, we use the Bayesian hierarchical modeling technique to incorporate prior information from similar patients. We use HBDM to re-analyze the Motor Activity Log (MAL) data of participants with chronic stroke included in two clinical trials: (1) the DOSE trial, in which participants were assigned to a 0, 15, 30, or 60-h dose condition (data of 40 participants analyzed), and (2) the EXCITE trial, in which participants were assigned a 60-h dose, in either an immediate or a delayed condition (95 participants analyzed).ResultsFor both datasets, HBDM accounts well for individual dynamics in the MAL during and outside of training: mean RMSE = 0.28 for all 40 DOSE participants (participant-level RMSE 0.26 ± 0.19—95% CI) and mean RMSE = 0.325 for all 95 EXCITE participants (participant-level RMSE 0.32 ± 0.31), which are small compared to the 0-5 range of the MAL. Bayesian leave-one-out cross-validation shows that the model has better predictive accuracy than static regression models and simpler dynamic models that do not account for the effect of supervised training, self-training, or forgetting. We then showcase model’s ability to forecast the MAL of “new” participants up to 8 months ahead. The mean RMSE at 6 months post-training was 1.36 using only the baseline MAL and then decreased to 0.91, 0.79, and 0.69 (respectively) with the MAL following the 1st, 2nd, and 3rd bouts of training. In addition, hierarchical modeling improves prediction for a patient early in training. Finally, we verify that this model, despite its simplicity, can reproduce previous findings of the DOSE trial on the efficiency, efficacy, and retention of motor therapy.ConclusionsIn future work, such forecasting models can be used to simulate different stages of recovery, dosages, and training schedules to optimize rehabilitation for each person.Trial registration This study contains a re-analysis of data from the DOSE clinical trial ID NCT01749358 and the EXCITE clinical trial ID NCT00057018

Multimodal preference heterogeneity in choice-based conjoint analysis: a simulation study

The most commonly used variant of conjoint analysis is choice-based conjoint (CBC). Here, hierarchical Bayesian (HB) multinomial logit (MNL) models are widely used for preference estimation at the individual respondent level. A new and very flexible approach to address multimodal and skewed preference heterogeneity in the context of CBC is the Dirichlet Process Mixture (DPM) MNL model. The number and masses of components do not have to be predisposed like in the latent class (LC) MNL model or in the mixture-of-normals (MoN) MNL model. The aim of this Monte Carlo study is to evaluate the performance of Bayesian choice models (basic MNL, HB-MNL, MoN-MNL, LC-MNL and DPM-MNL models) under varying data conditions (especially under multimodal heterogeneity structures) using statistical criteria for parameter recovery, goodness-of-fit and predictive accuracy. The core finding from this Monte Carlo study is that the standard HB-MNL model appears to be highly robust in multimodal preference settings.

Examining the Effect of Beech Bark Disease on Radial Growth in American Beech

Beech bark disease (BBD) is a significant threat to forests of North America and the impact of BBD on radial growth in the American beech is substantial. We developed a novel hierarchical Bayesian (HB) model to simultaneously model disease dynamics, tree growth, and the interaction of the two. Our model can be adapted to both emerging and more mature forest–pathogen systems to aid in ecosystem loss predictions. Long-term data from a single site minimized potential confounding variables such as climate change, precipitation, land use history, and soil conditions that may influence radial growth. Here, 206 beech trees were monitored over 15 years at an 85-acre site in southwestern Vermont, measuring diameter at breast height (DBH) and progression of BBD. Our model allows us to accurately estimate error rates in disease severity estimation and DBH measurements, and estimate the true state based on environmental variables. As disease poses significant threats to many tree species around the world, researchers can obtain more value and information from their datasets utilizing an adapted HB model.

Poisson-Lognormal model with measurement error in covariate for small area estimation of count data

Small Area Estimation (SAE) is a statistical technique to estimate parameters of subpopulation containing small-size of samples. SAE is an indirect estimation method by utilizing the strength of the neighbor area and data sources outside the area so that the sample becomes more effective and reduces the variance of parameter estimators. This paper deals with non-symmetrical count data in SAE which can be modelled based on Poisson-Lognormal distribution using hierarchical Bayesian (HB) approach and information on covariate contains measurement error. We develop the Poisson-Lognormal model for predicting small area counts with structural measurement error in the area-specific covariate. To get the HB Bayes estimator for Poisson-Lognormal model with measurement error in its covariates, the Metropolis-Hastings (MH) algorithm from the Markov Chain Monte Carlo (MCMC) technique is used. The HB estimator performance is studied through simulation and implementation of real data to predict the illiteracy rate at the sub-district level in Kepulauan Riau Province Indonesia based on The National Socio-Economic Survey (Susenas) data in March 2020.

The Hubble constant from two sibling Type Ia supernovae in the nearby galaxy NGC 4414: SN 1974G and SN 2021J

Having two “sibling” Type Ia supernovae (SNe Ia) in the same galaxy offers additional advantages in reducing a variety of systematic errors involved in estimating the Hubble constant, H0. NGC 4414 is a nearby galaxy included in the Hubble Space Telescope Key Project to measure its distance using Cepheid variables. It hosts two sibling SNe Ia: SN 2021J and SN 1974G. This provides the opportunity to improve the precision of the previous estimate of H0, which was based solely on SN 1974G. Here we present new optical BVRI photometry obtained at the Observatorio de Sierra Nevada and complement it with Swift UVOT UBV data, which cover the first 70 days of emission of SN 2021J. A first look at SN 2021J optical spectra obtained with the Gran Telescopio Canarias (GTC) reveals typical SN type Ia features. The main SN luminosity parameters for the two sibling SNe are obtained by using SNooPy, a light curve fitting code based on templates. Using a hierarchical bayesian approach, we build the Hubble diagram with a sample of 96 SNe Ia obtained from the Combined Pantheon Sample in the redshift range z = 0.02 − 0.075, and calibrate the zero point with the two sibling type-Ia SNe in NGC 4414. We report a value of the Hubble constant H0 = 72.19 ± 2.32 (stat.) ±3.42 (syst.) km s−1 Mpc−1. We expect a reduction of the systematic error after a new analysis of the Cepheids period-luminosity relation using the upcoming Gaia DR4 and additional Cepheids from the HST and JWST.

Forecasted 2040 global prevalence of nonalcoholic fatty liver disease using hierarchical bayesian approach.

Due to increases in obesity and type 2 diabetes, the prevalence of nonalcoholic fatty liver disease (NAFLD) has also been increasing. Current forecast models may not include non-obese NAFLD. Here, we used the Bayesian approach to forecast the prevalence of NAFLD through the year 2040. Prevalence data from 245 articles involving 2,699,627 persons were used with a hierarchical Bayesian approach to forecast the prevalence of NAFLD through 2040. Subgroup analyses were performed for age, gender, presence of metabolic syndrome, region, and smoking status. Sensitivity analysis was conducted for clinical setting and study quality. The forecasted 2040 prevalence was 55.7%, a three-fold increase since 1990 and a 43.2% increase from the 2020 prevalence of 38.9%. The estimated average yearly increase since 2020 was 2.16%. For those aged <50 years and ≥50 years, the 2040 prevalence were not significantly different (56.7% vs. 61.5%, P=0.52). There was a significant difference in 2040 prevalence by sex (males: 60% vs. 50%) but the trend was steeper for females (annual percentage change: 2.5% vs. 1.5%, P=0.025). There was no difference in trends overtime by region (P=0.48). The increase rate was significantly higher in those without metabolic syndrome (3.8% vs. 0.84%, P=0.003) and smokers (1.4% vs. 1.1%, P=0.011). There was no difference by clinical/community setting (P=0.491) or study quality (P=0.85). By 2040, over half the adult population is forecasted to have NAFLD. The largest increases are expected to occur in women, smokers, and those without metabolic syndrome. Intensified efforts are needed to raise awareness of NAFLD and to determine long-term solutions addressing the driving factors of the disease.

Estimation of waste‐to‐energy system reliability under hierarchical Bayesian

AbstractIn this paper, we discuss about the hierarchical Bayesian (HB) estimation concerning system reliability for a waste‐to‐energy (WTE) process. The main goal of this approach is to obtain this estimation WTE process reliability under different loss functions related series, parallel, and k‐out‐of‐m systems. In this case, we can drive system outcome by estimating each individual component. It is assumed that components are independent and identically distributed exponential random variables. Properties of the HB estimations under different loss functions are also provided, and comparisons are made between Bayesian and HB estimators via Monte Carlo simulation. The implementation of the proposed procedure was illustrated in detail by employing numerical practical examples extracted from a WTE process at the final part of this paper.

Hierarchical Bayesian Small Area Estimation Using Weakly Informative Priors in Ecologically Homogeneous Areas of the Interior Western Forests

The U.S. Forest Inventory and Analysis Program (FIA) collects inventory data on and computes estimates for many forest attributes to monitor the status and trends of the nation's forests. Increasingly, FIA needs to produce estimates in small geographic and temporal regions. In this application, we implement area level hierarchical Bayesian (HB) small area estimators of several forest attributes for ecosubsections in the Interior West of the US. We use a remotely-sensed auxiliary variable, percent tree canopy cover, to predict response variables derived from ground-collected data such as basal area, biomass, tree count, and volume. We implement four area level HB estimators that borrow strength across ecological provinces and sections and consider prior information on the between-area variation of the response variables. We compare the performance of these HB estimators to the area level empirical best linear unbiased prediction (EBLUP) estimator and to the industry-standard post-stratified (PS) direct estimator. Results suggest that when borrowing strength to areas which are believed to be homogeneous (such as the ecosection level) and a weakly informative prior distribution is placed on the between-area variation parameter, we can reduce variance substantially compared the analogous EBLUP estimator and the PS estimator. Explorations of bias introduced with the HB estimators through comparison with the PS estimator indicates little to no addition of bias. These results illustrate the applicability and benefit of performing small area estimation of forest attributes in a HB framework, as they allow for more precise inference at the ecosubsection level.

Simple and Two-Level Hierarchical Bayesian Approaches for Parameter Estimation with One- and Two-Layer Evapotranspiration Models of Crop Fields

The two-layer Shuttleworth–Wallace (SW) evapotranspiration (ET) model has been widely used for predicting ET with good results. Since the SW model has a large number of specific parameters, these parameters have been estimated using a simple non-hierarchical Bayesian (SB) approach. To further improve the performance of the SW model, we aimed to assess parameter estimation using a two-level hierarchical Bayesian (HB) approach that takes into account the variation in observed conditions through the comparison with a traditional one-layer Penman–Monteith (PM) model. The difference between the SB and HB approaches were evaluated using a field-based ET dataset collected from five agricultural fields over three seasons in Myanmar. For a calibration period with large variation in environmental factors, the models with parameters calibrated by the HB approach showed better fitting to observed ET than that with parameters estimated using the SB approach, indicating the potential importance of accounting for seasonal fluctuations and variation in crop growth stages. The validation of parameter estimation showed that the ET estimation of the SW model with calibrated parameters was superior to that of the PM model, and the SW model provided acceptable estimations of ET, with little difference between the SB and HB approaches.

Tree manipulation experiment for the short-term effect of tree cutting on N2O emission: A evaluation using Bayesian hierarchical modeling

Considerable uncertainty exists with regard to the effects of thinning and harvesting on N2O emissions as a result of changes caused in the belowground environment by tree cutting. To evaluate on the effects of changes in the belowground environment on N2O emissions from soils, we conducted a tree manipulation experiment in a Japanese cedar (Cryptomeria japonica) stand without soil compaction or slash falling near measurement chambers and measured N2O emission at distances of 50 and 150 cm from the tree stem (stump) before and after cutting. In addition, we inferred the effects of logging on the emission using a hierarchical Bayesian (HB) model. Our results showed that tree cutting stimulated N2O emission from soil and that the increase in N2O emission depended on the distance from the stem (stump); increase in N2O emission was greater at 50 than at 150 cm from the stem. Tree cutting caused the estimated N2O emission at 0–40 cm from the stem to double (the % increase in N2O emission by tree cutting was 54%–213%, 95% predictive credible interval) when soil temperature was 25 °C and WFPS was 60%. Posterior simulation of the HB model predicted that 30% logging would cause a 57% (47%–67%) increase in N2O emission at our study site (2000 trees ha−1) considering only the effects of belowground changes by tree cutting during the measurement period.

Estimating left behind patterns in congested metro systems: a Bayesian model

Purpose During rush hours, many passengers find it difficult to board the first train due to the insufficient capacity of metro vehicles, namely, left behind phenomenon. In this paper, a data-driven approach is presented to estimate left-behind patterns using automatic fare collection (AFC) data and train timetable data. Design/methodology/approach First, a data preprocessing method is introduced to obtain the waiting time of passengers at the target station. Second, a hierarchical Bayesian (HB) model is proposed to describe the left behind phenomenon, in which the waiting time is expressed as a Gaussian mixture model. Then a sampling algorithm based on Markov Chain Monte Carlo (MCMC) is developed to estimate the parameters in the model. Third, a case of Beijing metro system is taken as an application of the proposed method. Findings The comparison result shows that the proposed method performs better in estimating left behind patterns than the existing Maximum Likelihood Estimation. Finally, three main reasons for left behind phenomenon are summarized to make relevant strategies for metro managers. Originality/value First, an HB model is constructed to describe the left behind phenomenon in a target station and in the target direction on the basis of AFC data and train timetable data. Second, a MCMC-based sampling method Metropolis–Hasting algorithm is proposed to estimate the model parameters and obtain the quantitative results of left behind patterns. Third, a case of Beijing metro is presented as an application to test the applicability and accuracy of the proposed method.

Hierarchical Bayes Conjoint Choice Models - Model Framework, Bayesian Inference, Model Selection, and Interpretation of Estimation Results

Choice-based conjoint (CBC) is nowadays the most widely used variant of conjoint analysis, a class of methods for measuring consumer preferences. The primary reason for the increasing dominance of the CBC approach over the last 35 years is that it closely mimics real choice behavior of consumers by asking respondents repeatedly to choose their preferred alternative from a set of several offered alternatives (choice sets). Within the framework of CBC analysis, the multinomial logit (MNL) model is the most frequently used discrete choice model due to the existence of closed form solutions for conditional choice probabilities. The popularity of CBC and the MNL model has grown even more since the introduction of hierarchical Bayesian (HB) estimation techniques that accommodate individual consumer heterogeneity in choice data, and which have now become state-of-the-art in marketing theory and practice. Still, researchers and practitioners have to make further decisions under this framework (CBC, MNL, HB estimation), such as how to represent preference heterogeneity. Here, using a normal distribution (and therefore a unimodal distribution) has become the standard approach in the marketing literature. However, the thin tails of the normal distribution suggest that the standard HB-MNL model should not be the “go-to” approach to approximate multimodal preference distributions, because individual preference patterns lying at the tails of the normal distribution (i.e., that do not fit well with the assumption of a unimodal distribution) tend to be shrunk to the population mean. This shrinkage, especially in multimodal data settings, could mask important information (e.g., new or different structures in the data). A mixture of normal distributions avoids this limited flexibility of the most simple continuous approach of assuming a unimodal prior heterogeneity distribution. There are currently two prominent HB-CBC modeling approaches embedding the mixture-of-normals (MoN) approach: the more widespread MoN-HB-MNL model, and the Dirichlet process mixture (DPM)-HB-MNL model. In this article, we review the prominent HB-MNL model (with its normal prior), the MoN-HB-MNL model, and the DPM-HB-MNL model and apply them to an empirical multi-country CBC data set. We compare the statistical performance of the three models in terms of goodness-of-fit and predictive accuracy, show how to include consumer background characteristics in the upper level of these models, and illustrate how to interpret the estimation results (with a special focus on cross-county heterogeneity). In sum, our article serves as a kind of user guide to the estimation and interpretation of Hierarchical Bayes Conjoint Choice Models. For our data, we observed that all three choice models (both with and without consumer background characteristics) resulted in a one-component solution. The DPM-HB-MNL model nevertheless yielded a higher cross-validated hit rate compared to the MoN-HB-MNL and the HB-MNL models due to its even more flexible prior assumptions. The two latter models tended to slightly overfit our empirical data, which was reflected by higher goodness-of-fit statistics but a lower predictive accuracy compared to the DPM-HB-MNL model. We showed that this result could be attributed to the weaker extent of Bayesian shrinkage of these two models. The DPM-HB-MNL model showed a stronger shrinkage effect and seems therefore somewhat more robust against overfitting. Including consumer background characteristics in terms of country of origin information for the respondents did not improve the statistical model performance (especially not the predictive performance). Still, using the country of origin information for respondents in a post-hoc segmentation analysis helped us to explain some differences in brand preferences between the five countries.

Modeling of moral decisions with deep learning

One example of an artificial intelligence ethical dilemma is the autonomous vehicle situation presented by Massachusetts Institute of Technology researchers in the Moral Machine Experiment. To solve such dilemmas, the MIT researchers used a classic statistical method known as the hierarchical Bayesian (HB) model. This paper builds upon previous work for modeling moral decision making, applies a deep learning method to learn human ethics in this context, and compares it to the HB approach. These methods were tested to predict moral decisions of simulated populations of Moral Machine participants. Overall, test results indicate that deep neural networks can be effective in learning the group morality of a population through observation, and outperform the Bayesian model in the cases of model mismatches.

Choice data generation using usage scenarios and discounted cash flow analysis

Discrete choice analysis is a popular method of estimating heterogeneous customer preferences. Although model accuracy can be increased by including more choice data, this option is untenable when the obtaining of choice data from target customers is costly and time-consuming.. We thus propose a method for choice data generation for commercial products whose expected money value is a key factor in consumer choice (e.g., commercial vehicles and financial product). Using an individual usage scenario, we generate a discounted cash flow (DCF) model instead of a utility model to estimate the discount rates, rather than partworths, of individual consumers. The DCF model helps us generate synthetic choice data from choice sets consisting of various combinations of attribute levels. Using these data, we employ a hierarchical Bayesian (HB) discrete choice analysis. We conclude the study with a case study on the preference estimation of a hybrid courier truck conversion. The results reveal that the DCF-based HB estimation outperforms the traditional HB estimation.

On hierarchical bayesian based predictive maintenance of autonomous natural gas regulating operations

Safety Improvement of engineering processes, especially Oil & Gas operations, has gained a lot of attention during the last decades. This fundamental vision results in risk remediation programs, minimizing the risks of failure, and reducing the associated costs for operation and maintenance. As failures may represent serious threats for both humans and the environment, a comprehensive tool is required to employ maintenance and avoid immoderate dangerous consequences. Traditional risk frameworks mainly include estimation approaches, such as Fault Tree (FT) and Event Tree (ET), producing more simplified models than other tools, such as Bayesian inference. The present work aimed at developing an advanced Risk-Based Maintenance (RBM) methodology for prioritizing maintenance operations, by addressing associated uncertainties through the accident modelling of the process. For this purpose, a Hierarchical Bayesian Approach (HBA) is applied to estimate the failure probabilities of each component while a Failure Mode, Effects and Criticality Analysis is performed to assess the severity. With Markov Chain Monte Carlo simulation from likelihood function and prior distribution, the HBA is capable of incorporating the fluctuations and uncertainties associated with operational data including the variability between the source of data and the correlation of observations. Lastly, to make a meaningful difference between different kinds of risk consequences, whether the risk has a direct or indirect loss, the cost of failures of components is accounted for. To demonstrate the application of the methodology, a Natural Gas Reduction and Measuring Station (NGRMS) is taken into account as a case study. The outcome of the case study proofed that PTG and pump are the most failures sensitive components among other if they being left unattended in the operation with an average number of failure occurrences of 67 and 45; While the THT pipelines and THT tank are less sensitive for being considered for major maintenance request with almost average of 5 times in their lifetime. The proposed method can be exploited by maintenance engineers, asset managers, and policymakers to reduce the downtime periods as well as the risks of on-going operations.