Loss Of Prediction Accuracy Research Articles

Accurate Loss Prediction of Realistic Hollow-Core Anti-Resonant Fibers Using Machine Learning

Accurate Loss Prediction of Realistic Hollow-Core Anti-Resonant Fibers Using Machine Learning

Multiple fairness criteria in decision tree learning

The use of algorithmic decision-making systems based on machine learning models has led to a need for fair (unbiased) and explainable classification outcomes. In particular, machine learning algorithms can encode biases, which might result in discriminatory decisions for certain groups such as gender, race, or age. Although a number of works on decision tree learning have been proposed to decrease the chance of discrimination, they usually focus on the use of a single fairness metric. In general, creating a model based on a single fairness metric is not a sufficient way to mitigate discrimination since bias can originate from various sources—e.g., the data itself or the optimization process. In this paper, we propose a novel decision tree learning process that utilizes multiple fairness metrics to address both group and individual discrimination. This is achieved by extending the attribute selection procedure to consider not only information gain but also gain in fairness. Computational experiments on fourteen different datasets with various sensitive features demonstrate that the proposed Fair-C4.5 models improve fairness without a loss in predictive accuracy when compared to the well-known C4.5 and the fairness-aware FFTree algorithms.

The Impact of the Weather Forecast Model on Improving AI-Based Power Generation Predictions through BiLSTM Networks

This study aims to comprehensively analyze five weather forecasting models obtained from the Open-Meteo historical data repository, with a specific emphasis on evaluating their impact in predicting wind power generation. Given the increasing focus on renewable energy, namely, wind power, accurate weather forecasting plays a crucial role in optimizing energy generation and ensuring the stability of the power system. The analysis conducted in this study incorporates a range of models, namely, ICOsahedral Nonhydrostatic (ICON), the Global Environmental Multiscale Model (GEM Global), Meteo France, the Global Forecast System (GSF Global), and the Best Match technique. The Best Match approach is a distinctive solution available from the weather forecast provider that combines the data from all available models to generate the most precise forecast for a particular area. The performance of these models was evaluated using various important metrics, including the mean squared error, the root mean squared error, the mean absolute error, the mean absolute percentage error, the coefficient of determination, and the normalized mean absolute error. The weather forecast model output was used as an essential input for the power generation prediction models during the evaluation process. This method was confirmed by comparing the predictions of these models with actual data on wind power generation. The ICON model, for example, outscored others with a root mean squared error of 1.7565, which is a tiny but essential improvement over Best Match, which had a root mean squared error of 1.7604. GEM Global and Gsf Global showed more dramatic changes, with root mean squared errors (RMSEs) of 2.0086 and 2.0242, respectively, indicating a loss in prediction accuracy of around 24% to 31% compared to ICON. Our findings reveal significant disparities in the precision of the various models used, and certain models exhibited significantly higher predictive precision.

Deviation entropy-based dynamic multi-model ensemble interval prediction method for quantifying uncertainty of building cooling load

Interval prediction is a promising method that can reveal the uncertainty of building load and has been shown to effectively manage building energy systems. Previous studies focused on improving the performance of three types of generic single models (quantile regression, bootstrap, and lower upper bound estimation models). Multi-model ensemble methods have the advantages of reducing overfitting risk, improving stability. However, in the field of building cooling load interval prediction, there is a research gap that the performance of the multi-model ensemble method has not been conducted extensively. To compensate for this gap, firstly, referring to the information entropy in information theory, we proposed deviation entropy, which is a new concept that can be used to calculate the weight of a generic single model. Based on this, a static multi-model ensemble interval prediction method was developed. Then, we introduced a sliding time window to further upgrade the static method into a dynamic method. Finally, we used actual data to evaluate the performance of the proposed dynamic multi-model ensemble interval prediction method. The study results show that the multi-model ensemble method outperforms traditional generic single model, and the dynamic method can improve the prediction reliability by 8.5% with only 0.76% loss of prediction accuracy.

Evaluation of electric vehicle response capability in power grid

In recent years, frequent and substantial area-wide power outages have underscored the critical need for cities to possess robust backup power sources capable of swift response to prevent prolonged power system disruptions. Electric vehicles can contribute electricity to the power grid using vehicle-to-grid technology. The power delivered by electric vehicles in this context is termed as response capability. However, existing studies have overlooked response capability dynamics during transitions between electric vehicle states—such as the shift from charging or discharging to an idle state, thereby hindering a comprehensive understanding of this aspect. Hence, this paper introduces a multi-timescale response capability prediction model that evaluates the electric vehicle's state of charge to ensure users’ requirements are met for upcoming trips. To better assess users' travel demand, the gravity model is employed as a precursor to response capability prediction to further enhance the validity of the prediction outcomes. Three neighborhoods in Los Angeles have been chosen for analysis: Downtown, Lincoln Heights, and Silver Lake. Predictions indicate that neglecting the response capability when electric vehicles undergo state transformation can lead to a differential response capability ranging from 2000 kWh to 4000 kWh, resulting in a loss of prediction accuracy by 20 % to 25 %.•The response capability of EV is non-zero during state transformations•Users’ travel demand assessment•Seamless integration of vehicle-to-grid technology into the power grid.

Wasserstein Robust Classification with Fairness Constraints

Problem definition: Data analytics models and machine learning algorithms are increasingly deployed to support consequential decision-making processes, from deciding which applicants will receive job offers and loans to university enrollments and medical interventions. However, recent studies show these models may unintentionally amplify human bias and yield significant unfavorable decisions to specific groups. Methodology/results: We propose a distributionally robust classification model with a fairness constraint that encourages the classifier to be fair in the equality of opportunity criterion. We use a type-[Formula: see text] Wasserstein ambiguity set centered at the empirical distribution to represent distributional uncertainty and derive a conservative reformulation for the worst-case equal opportunity unfairness measure. We show that the model is equivalent to a mixed binary conic optimization problem, which standard off-the-shelf solvers can solve. We propose a convex, hinge-loss-based model for large problem instances whose reformulation does not incur binary variables to improve scalability. Moreover, we also consider the distributionally robust learning problem with a generic ground transportation cost to hedge against the label and sensitive attribute uncertainties. We numerically examine the performance of our proposed models on five real-world data sets related to individual analysis. Compared with the state-of-the-art methods, our proposed approaches significantly improve fairness with negligible loss of predictive accuracy in the testing data set. Managerial implications: Our paper raises awareness that bias may arise when predictive models are used in service and operations. It generally comes from human bias, for example, imbalanced data collection or low sample sizes, and is further amplified by algorithms. Incorporating fairness constraints and the distributionally robust optimization (DRO) scheme is a powerful way to alleviate algorithmic biases. Funding: This work was supported by the National Science Foundation [Grants 2342505 and 2343869] and the Chinese University of Hong Kong [Grant 4055191]. Supplemental Material: The online appendices are available at https://doi.org/10.1287/msom.2022.0230 .

Revolutionizing power grid loss prediction with advanced hybrid time series deep learning model

Accurate prediction of grid loss in power distribution networks is pivotal for efficient energy management and pricing strategies. Traditional forecasting approaches often struggle to capture the complex temporal dynamics and external influences inherent in grid loss data. In response, this research presents a novel hybrid time-series deep learning model: Gated Recurrent Units with Temporal Convolutional Networks (GRU-TCN), designed to enhance grid loss prediction accuracy. The proposed model integrates the temporal sensitivity of GRU with the local context awareness of TCN, exploiting their complementary strengths. A learnable attention mechanism fuses the outputs of both architectures, enabling the model to discern significant features for accurate prediction. The model is evaluated using well-established metrics across distinct temporal phases: training, testing, and future projection. Results showcase Resulting in encouraging Figures for mean absolute error, root mean squared error, and mean absolute percentage error, the model’s capacity to capture both long-term trends and transitory patterns. The GRU-TCN hybrid model represents a pioneering approach to power grid loss prediction, offering a flexible and precise tool for energy management. This research not only advances predictive accuracy but also lays the foundation for a smarter and more sustainable energy ecosystem, poised to transform the landscape of energy forecasting.

Crop Canopy Nitrogen Estimation from Mixed Pixels in Agricultural Lands Using Imaging Spectroscopy

Accurate retrieval of canopy nutrient content has been made possible using visible-to-shortwave infrared (VSWIR) imaging spectroscopy. While this strategy has often been tested on closed green plant canopies, little is known about how nutrient content estimates perform when applied to pixels not dominated by photosynthetic vegetation (PV). In such cases, contributions of bare soil (BS) and non-photosynthetic vegetation (NPV), may significantly and nonlinearly reduce the spectral features relied upon for nutrient content retrieval. We attempted to define the loss of prediction accuracy under reduced PV fractional cover levels. To do so, we utilized VSWIR imaging spectroscopy data from the Global Airborne Observatory (GAO) and a large collection of lab-calibrated field samples of nitrogen (N) content collected across numerous crop species grown in several farming regions of the United States. Fractional cover values of PV, NPV, and BS were estimated from the GAO data using the Automated Monte Carlo Unmixing algorithm (AutoMCU). Errors in prediction from a partial least squares N model applied to the spectral data were examined in relation to the fractional cover of the unmixed components. We found that the most important factor in the accuracy of the partial least squares regression (PLSR) model is the fraction of photosynthetic vegetation (PV) cover, with pixels greater than 60% cover performing at the optimal level, where the coefficient of determination (R2) peaks to 0.66 for PV fractions of more than 60% and bare soil (BS) fractions of less than 20%. Our findings guide future spaceborne imaging spectroscopy missions as applied to agricultural cropland N monitoring.

Best performances of visible–near-infrared models in soils with little carbonate – a field study in Switzerland

Abstract. Conventional laboratory analysis of soil properties is often expensive and requires much time if various soil properties are to be measured. Visual and near-infrared (vis–NIR) spectroscopy offers a complementary and cost-efficient way to gain a wide variety of soil information at high spatial and temporal resolutions. Yet, applying vis–NIR spectroscopy requires confidence in the prediction accuracy of the infrared models. In this study, we used soil data from six agricultural fields in eastern Switzerland and calibrated (i) field-specific (local) models and (ii) general models (combining all fields) for soil organic carbon (SOC), permanganate oxidizable carbon (POXC), total nitrogen (N), total carbon (C) and pH using partial least-squares regression. The 30 local models showed a ratio of performance to deviation (RPD) between 1.14 and 5.27, and the root mean square errors (RMSE) were between 1.07 and 2.43 g kg−1 for SOC, between 0.03 and 0.07 g kg−1 for POXC, between 0.09 and 0.14 g kg−1 for total N, between 1.29 and 2.63 g kg−1 for total C, and between 0.04 and 0.19 for pH. Two fields with high carbonate content and poor correlation between the target properties were responsible for six local models with a low performance (RPD < 2). Analysis of variable importance in projection, as well as of correlations between spectral variables and target soil properties, confirmed that high carbonate content masked absorption features for SOC. Field sites with low carbonate content can be combined with general models with only a limited loss in prediction accuracy compared to the field-specific models. On the other hand, for fields with high carbonate contents, the prediction accuracy substantially decreased in general models. Whether the combination of soils with high carbonate contents in one prediction model leads to satisfying prediction accuracies needs further investigation.

대용량 자료의 이항 분류에서의 충분 차원 축소를 위한 전향적 접근 방법

Sufficient dimension reduction, aimed at finding a lower-dimensional subspace in explanatory variables that contains response variable information, typically relies on inverse-based methodologies. These methods are easy to implement but often require linear or constant variance conditions. To address these limitations, techniques for forwardly estimating the central subspace have been developed. In particular, methods utilizing the Reproducing Kernel Hilbert Space have gained attention, but their use in analyzing large datasets is limited due to the characteristics of the kernel space. In this paper, we study a novel forward approach for sufficient dimension reduction in binomial classification of large-scale data. We propose a method that employs a divide-and-conquer technique to split data into subsets, then independently perform dimension reduction on each subset before synthesizing them into a final model. It was shown that when the number of partitions of data was appropriately selected, the loss in prediction accuracy was not significant compared to the existing method, while being efficient in terms of storage space and calculation cost. In addition, simulations in various models showed superior prediction accuracy than other inverse-based techniques. The utility of the proposed method was confirmed through the real data analysis.

Multistep traffic speed prediction: A sequence-to-sequence spatio-temporal attention model

Multistep traffic speed prediction plays a crucial role in alleviating road congestion and improving transport efficiency. In actual traffic networks, the spatio-temporal dependence among roads dynamically changes over time due to factors such as road conditions and unforeseen incidents, which brings great challenges to multistep traffic speed prediction. Additionally, multistep traffic speed prediction commonly faces the problem of error accumulation, resulting in a loss of prediction accuracy. To address these issues, we propose a Sequence-to-Sequence Spatio-Temporal Attention model (STSSTA) for multistep traffic speed prediction. Specifically, we build an adaptive tuning module to select road preferences and automatically obtain global information about the roads. Then we construct a Sequence-to-Sequence architecture for spatio-temporal feature learning and multistep traffic speed prediction. In particular, in the encoder, we design a Diffusion Graph Convolutional Network (DGCN) and combine it with a Gated Recurrent Unit (GRU) to effectively capture the complex spatio-temporal features within the traffic networks. In the decoder, we introduce a Recalling attention mechanism to alleviate the problem of local information loss caused by encoder compression, thereby reducing error accumulation in multistep prediction. Experiments on METR-LA and PeMS-BAY datasets demonstrate that STSSTA outperforms the baseline models in long-term prediction.

Optimization design of radial inflow turbine combined with mean-line model and CFD analysis for geothermal power generation

It is a considerable challenge to determine the key parameters affecting the efficiency and propose an accurate loss prediction model for radial flow turbine design. In current investigation, a one-dimensional mean-line model combined with particle swarm optimization algorithm and Kriging response surface surrogate model based on 3D numerical results were proposed to optimize radial flow turbines and evaluate performance. The loss model was carried out to analyze the relationship between geometric parameters, operating conditions and turbine performance. CFD numerical simulations were employed to revealed the mechanism of enhancing the turbine performance. The total-static efficiency was increased from 88.5 % to 91.7 %, and the clearance loss and passage loss were reduced by 18.4 % and 35.8 %, respectively,by optimized design. It is attributed to mitigate the curvature-induced boundary layer separation and vortex at the outlet, which reduces the secondary flow losses.The correlation between the predicted values of the Kriging response surface and the numerical results was up to 98.3 %. The results showed that the angle of attack was in the range of -10-0°, the blade has less influence on the flow loss. The current study effectively combined the flow path structure parameters and flow characteristics to realize the efficient optimization of ORC turbine.

Assessing the potential of remote sensing-based models to predict old-growth forests on large spatiotemporal scales

Old-growth forests provide a broad range of ecosystem services. However, due to poor knowledge of their spatiotemporal distribution, implementing conservation and restoration strategies is challenging. The goal of this study is to compare the predictive ability of socioecological factors and different sources of remotely sensed data that determine the spatiotemporal scales at which forest maturity attributes can be predicted.We evaluated various remotely sensed data that cover a broad range of spatial (from local to global) and temporal (from current to decades) extents, from Airborne Laser Scanning (ALS), aerial multispectral and stereo-imagery, Sentinel-1, Sentinel-2 and Landsat data. Using random forests, remotely sensed data were related to a forest maturity index available in 688 forest plots across four ranges of the French Alps. Each model also includes socioecological predictors related to topography, socioeconomy, pedology and climatology.We found that the different remotely sensed data provide information on the main forest structural characteristics as defined by ALS, except for Landsat, which has a too coarse resolution, and Sentinel-1, which responds differently to vegetation structure. The predictions were quite similar considering aerial remotely sensed data, on the one hand, and satellite remotely sensed data, on the other hand. Socioecological variables are the most important predictors compared to the remote sensing metrics.In conclusion, our results indicate that a wide range of remotely sensed data can be used to study old-growth forests beyond the use of ALS and despite different abilities to predict forest structure. Accounting for socioecological predictors is indispensable to avoid a significant loss of predictive accuracy. Remotely sensed data can allow for predictions to be made at different spatiotemporal resolutions and extents. This study paves the way to large-scale monitoring of forest maturity, as well as for retrospective analyses which will show to what extent predicted maturity change at different dates.

PRISM: A Hierarchical Intrusion Detection Architecture for Large-Scale Cyber Networks

The increase in scale of cyber networks and the rise in sophistication of cyber-attacks have introduced several challenges in intrusion detection. The primary challenge is the requirement to detect complex multi-stage attacks in realtime by processing the immense amount of traffic produced by present-day networks. In this paper we present PRISM, a hierarchical intrusion detection architecture that uses a novel attacker behavior model-based sampling technique to minimize the realtime traffic processing overhead. PRISM has a unique multi-layered architecture that monitors network traffic distributedly to provide efficiency in processing and modularity in design. PRISM employs a Hidden Markov Model-based prediction mechanism to identify multi-stage attacks and ascertain the attack progression for a proactive response. Furthermore, PRISM introduces a stream management procedure that rectifies the issue of alert reordering when collected from distributed alert reporting systems. To evaluate the performance of PRISM, multiple metrics have been proposed, and various experiments have been conducted on a multi-stage attack dataset. The results exhibit up to 7.5x improvement in processing overhead as compared to a standard centralized IDS without the loss of prediction accuracy while demonstrating the ability to predict different attack stages promptly.

Prediction of time-varying dynamics and chatter stability analysis for surface milling of thin-walled curved CFRP workpiece

Machining curved surfaces for thin-walled components is always challenging. Due to the strong anisotropic features of carbon fiber reinforced plastic (CFRP), new challenges, such as prediction accuracy loss, have emerged for avoiding the chatter of machining thin-walled curved CFRP workpieces. This paper studied the influence of the anisotropic properties of CFRP workpieces and time-varying dynamics on surface milling stability. An improved damping and time-varying dynamics prediction model was proposed integrated with FEA and numerical calculation considering the change of cutting position, the materials removed, and the anisotropic features of CFRP. Then, the 3D stability lobe diagram (SLD) was achieved by considering the dynamic change at each cutter point on the machining path and dynamic cutting forces based on the instantaneous fiber cutting angle. Based on the obtained 3D SLD, CFRP milling chatter could be accurately avoided, and machining efficiency could be improved. A case study of milling a thin-walled toroidal surface was conducted for validation. In the machining process of CFRP, unlike homogeneous materials such as metals, the natural frequency of CFRP fluctuates nonlinearly with the change curve of material removal within a specific range. Under a higher material removal rate, the surface roughness of the machined surface has been reduced by 74%, Ra below 2.35 µm, based on the selected process parameters of the novel-established SLD, compared to the surface roughness, Ra close to 9.21 µm, obtained by the previous method without considering the time-varying dynamics.

Epistemic modeling uncertainty of rapid neural network ensembles for adaptive learning

Emulator embedded neural networks, which are closely related to physics informed neural networks, leverage multi-fidelity data sources for efficient design exploration of aerospace engineering systems. Multiple realizations of the neural network model are trained with different random initializations. The ensemble of model realizations is used to assess epistemic modeling uncertainty caused by a lack of training samples. This uncertainty estimation is crucial information for successful goal-oriented adaptive learning in an aerospace system design exploration. However, the costs of training the ensemble models often become prohibitive and pose a computational challenge, especially when the models are not trained in parallel during adaptive learning. In this work, a new type of emulator embedded neural network is presented using the rapid neural network paradigm. Unlike the conventional neural network training that optimizes the weights and biases of all the network layers by using gradient-based backpropagation, rapid neural network training adjusts only the last layer connection weights by applying a linear regression technique. It is found that the proposed emulator embedded neural network trains near-instantaneously, typically without loss of prediction accuracy. The proposed method is demonstrated on multiple analytical examples, as well as an aerospace flight parameter study of a generic hypersonic vehicle.

Neural net-based surrogate square pin fin correlations for thermal-fluid design optimization under developing flow conditions

Heat exchangers and cold plates are widely used for thermal management of electronic devices and in energy storage applications. Accurate prediction of the thermal-fluid transport performance of heat transfer surfaces is crucial for design optimization of these thermal management components. Microscale pin fin structures are commonly employed for enhancement of convection heat transfer. Previous studies have focused on fully developed flow through circular and square pin fins in staggered arrangements, leaving a gap in predictive correlations available for in-line arrangements of square pin fins with consideration of hydrodynamic and thermal developing effects. This study aims to investigate and develop robust friction factor (f) and Nusselt number (Nu) correlations for developing flow through in-line square microscale pin fin structures. Traditional empirical correlations are highly geometry dependent and assume functional forms that could introduce inaccuracies. On the other hand, accurate data may be available at specific points, but continuous evaluation of the function value and derivatives as needed for design optimization processes is difficult. To overcome these limitations, artificial neural network (ANN)-based surrogate models are developed to provide accurate and continuous-valued correlations. The proposed models, trained on numerically simulated data using water as the working fluid, provide accurate predictions of f and Nu, with a near exact match to the training data as well as on unseen testing data. Furthermore, the outputs of the ANNs are inspected to propose new analytical correlations to estimate the hydrodynamic and thermal entrance lengths for flow through square pin fin arrays. The ML models are also shown to be useable for fluids other than water, employing physics-based, Prandtl-number-dependent scaling relations. These generalized ML models have promise to significantly reduce computation cost in the design optimization of thermal management components having pin fin structures for enhanced convection, but without loss in prediction accuracy associated with alternative reduced-order correlations.

A data-driven flow loss prediction model for the blade hub region of a boundary layer ingestion fan rotor

PurposeIn a boundary layer ingestion (BLI) propulsion system, the fan operates continuously under distorted inflow conditions, leading to an increment of aerodynamic loss and in turn impacting the potential fuel burn reduction of the aircraft. Usually, in the preliminary design stage of a BLI propulsion system, it is essential to assess the impact of fuselage boundary layer fluids on fan aerodynamic performances under various flight conditions. However, the hub region flow loss is one of the major loss sources in a fan and would greatly influence the fan performances. Moreover, the inflow distortion also results in a complex and highly nonlinear mapping relation between loss and local physical parameters. It will diminish the prediction accuracy of the commonly used low-fidelity computational approaches which often incorporate traditional physics-based loss models, reducing the reliability of these approaches in evaluating fan performances. Meanwhile, the high-fidelity full-annulus unsteady Reynolds-averaged Navier–Stokes (URANS) approach, even though it can give rather accurate loss predictions, is extremely time-consuming. This study aims to develop a fast and accurate hub loss prediction method for a BLI fan under distorted inflow conditions.Design/methodology/approachThis paper develops a data-driven hub loss prediction method for a BLI fan under distorted inflows. To improve the prediction accuracy and applicability, physical understandings of hub flow features are integrated into the modeling process. Then, the key physical parameters related to flow loss are screened by conducting a sensitivity analysis of influencing parameters. Next, a quasi-steady assumption of flow is made to generate a training sample database, reducing the computational time by acquiring one single sample from the highly time-consuming full-annulus URANS approach to a cost-efficient single-blade-passage approach. Finally, a radial basis function neural network is used to establish a surrogate model that correlates the input parameters and the output loss.FindingsThe data-driven hub loss model shows higher prediction accuracy than the traditional physics-based loss models. It can accurately capture the circumferentially and radially nonuniform variation trends of the losses and the associated absolute magnitudes in a BLI fan under different blade load, inlet distortion intensity and rotating speed conditions. Compared with the high-fidelity full-annulus URANS results, the averaged relative prediction errors of the data-driven hub loss model are kept less than 10%.Originality/valueThe originality of this paper lies in developing a new method for predicting flow loss in a BLI fan rotor blade hub region. This method offers higher prediction accuracy than the traditional loss models and lower computational time cost than the full-annulus URANS approach, which could realize fast evaluations of fan aerodynamic performances and provide technical support for designing high-performance BLI fans.

Enabling Equal Opportunity in Logistic Regression Algorithm

Research Question: This paper aims at adjusting the logistic regression algorithm to mitigate unwanted discrimination shown towards race, gender, etc. Motivation: Decades of research in the field of algorithm design have been dedicated to making a better prediction model. Many algorithms are designed and improved, which made them better than the judgments of people and even experts. However, in recent years it has been discovered that predictive models can make unwanted discrimination. Such unwanted discrimination in the predictive model can lead to legal consequences. In order to mitigate the problem of unwanted discrimination, we propose equal opportunity between privileged and discriminated groups in the logistic regression algorithm. Idea: Our idea is to add a regularization term in the goal function of the logistic regression. Therefore, our predictive model will solve both the social problem and the predictive problem. More specifically, our model will provide fair and accurate predictions. Data: The data used in this research present U.S. census data describing individuals using personal characteristics with a goal to provide a binary classification model for predicting if an individual has an annual salary above $50k. The dataset used is known for disparate impact regarding female individuals. In addition, we used the COMPAS dataset aimed at predicting recidivism. COMPAS is biased toward African-Americans. Tools: We developed a novel regularization technique for equal opportunity in the logistic regression algorithm. The proposed regularization is compared against classical logistic regression and fairness constraint logistic regression, using a ten-fold cross-validation. Findings: The results suggest that equal opportunity logistic regression manages to create a fair prediction model. More specifically, our model improved both disparate impact and equal opportunity compared to classical logistic regression, with a minor loss in prediction accuracy. Compared to the disparate impact constrained logistic regression, our approach has higher prediction accuracy and equal opportunity, while having a lower disparate impact. By inspecting the coefficients of our approach and classical logistic regression, one can see that proxy attribute coefficients are reduced to very low values. Contribution: The main contribution of this paper is in the methodological part. More specifically, we implemented an equal opportunity in the logistic regression algorithm.

LASSO and Elastic Net Tend to Over-Select Features

Machine learning methods have been a standard approach to select features that are associated with an outcome and to build a prediction model when the number of candidate features is large. LASSO is one of the most popular approaches to this end. The LASSO approach selects features with large regression estimates, rather than based on statistical significance, that are associated with the outcome by imposing an L1-norm penalty to overcome the high dimensionality of the candidate features. As a result, LASSO may select insignificant features while possibly missing significant ones. Furthermore, from our experience, LASSO has been found to select too many features. By selecting features that are not associated with the outcome, we may have to spend more cost to collect and manage them in the future use of a fitted prediction model. Using the combination of L1- and L2-norm penalties, elastic net (EN) tends to select even more features than LASSO. The overly selected features that are not associated with the outcome act like white noise, so that the fitted prediction model may lose prediction accuracy. In this paper, we propose to use standard regression methods, without any penalizing approach, combined with a stepwise variable selection procedure to overcome these issues. Unlike LASSO and EN, this method selects features based on statistical significance. Through extensive simulations, we show that this maximum likelihood estimation-based method selects a very small number of features while maintaining a high prediction power, whereas LASSO and EN make a large number of false selections to result in loss of prediction accuracy. Contrary to LASSO and EN, the regression methods combined with a stepwise variable selection method is a standard statistical method, so that any biostatistician can use it to analyze high-dimensional data, even without advanced bioinformatics knowledge.