Uplift Model Research Articles

The Multi‐Segment Complexity of the 2024 MW ${M}_{W}$ 7.5 Noto Peninsula Earthquake Governs Tsunami Generation

AbstractThe 1 January 2024, moment magnitude 7.5 Noto Peninsula earthquake ruptured in complex ways, challenging analysis of its tsunami generation. We present tsunami models informed by a 6‐subevent centroid moment tensor (CMT) model obtained through Bayesian inversion of teleseismic and strong motion data. We identify two distinct bilateral rupture episodes. Initial, onshore rupture toward the southwest is followed by delayed re‐nucleation at the hypocenter, likely aided by fault weakening, causing significant seafloor uplift to the northeast. We construct a complex multi‐fault uplift model, validated against geodetic observations, that aligns with known fault system geometries and is critical in modeling the observed tsunami. The simulations can explain tsunami wave amplitude, timing, and polarity of the leading wave, which are crucial for tsunami early warning. Upon comparison with alternative source models and analysis of 2000 multi‐CMT ensemble solutions, we highlight the importance of incorporating complex source effects for realistic tsunami simulations.

High-elevation Qilian Mountains and its inspiration for tectonics and biodiversity during the late Middle Miocene

The accurate paleoelevation reconstruction of the Qilian Mountains is critical to advancing our understanding the integrity of the Tibetan Plateau (TP) uplift model, its deep structural mechanisms, and corresponding connections with climatic, environmental, and biodiversity changes. Recently, the first quantitative reconstruction of the paleomidrange (i.e., average elevation of the basin and mountains) of the northern Tibetan Plateau (NTP) was completed, using innovative palynological paleoaltimetry-TP/TPAP [(Tsuga% + Podocarpus%) / (Tsuga% + Podocarpus% + Abies% + Picea%)] ratios, which revealed a rapid uplift of the NTP from low to high elevations during the late Middle Miocene. Here, we analyzed the Ebotu Fauna pollen record (13–12 Ma) from the Hongyazi Basin, situated within the Qilian Mountains, to directly infer the paleoelevation of this region. The pollen assemblages were predominantly composed of conifers (average of 64.0 %), including Picea, Cedrus, and Pinus of the Pinaceae, as well as Cupressaceae, with broadleaves and steppes taxa each accounting for <15 %. This pollen composition evidently reflects a vegetation type dominated by high-mountain conifers forest, consistent with Middle Miocene pollen assemblages from the Qaidam Basin. Application TP/TPAP ratios yielded a paleomidrange of 3492 ± 87 m at 13–12 Ma. Combined with the known elevation of the Qaidam Basin during this period (1885 ± 566 m), the elevation of the Qilian Mountains was calculated to be 4338 ± 653 m. In turn, an elevation of 2646 ± 740 m was obtained for the Hongyazi Basin. This high-elevation terrain provides evidence to support the multi-stage convective removal model of the TP caused by crustal shortening and thickening. It also led to the formation of a humid ecosystem dominated by conifers forest in the Qilian Mountains, supporting the diversification of mammalian taxa.

Application of Advanced Machine Learning Models for Uplift and Penetration Resistance in Clay-Embedded Dual Interfering Pipelines

Application of Advanced Machine Learning Models for Uplift and Penetration Resistance in Clay-Embedded Dual Interfering Pipelines

Uplift modeling and its implications for appointment date prediction in attended home delivery

Successful attended home delivery (AHD) is the most important aspect of e-commerce order fulfillment. Prior literature focuses on incentive scheme development for customers' choices of delivery windows and predictive analytics for delivery results, but it is not clear whether the effect of AHD on the appointment date set by customers increases the success rate of AHD. Therefore, we developed an uplift modeling method, PSM-NDML, as a relevant prescriptive analytic tool for AHD on an appointment date, which aims to estimate the causal effect of the by-appointment delivery on the delivery result. PSM-NDML integrates propensity score matching and double machine learning, effectively addressing sample selection bias, low predictive performance, and poor interpretability. Applied to a real-world product delivery dataset of a Chinese logistics company, PSM-NDML achieves superior performance relative to ten other state-of-the-art uplift models in terms of cumulative gain and the Qini coefficient. The predicted responses gained from PSM-NDML are also visually interpreted at the global and local levels, which reveals various managerial insights. In practice, the findings expand managers' understanding of the heterogeneous effects of AHD on appointment dates and provide decision support for logistics companies in the development of home delivery plans.

Improving incentive policies to salespeople cross-sells: a cost-sensitive uplift modeling approach

Improving incentive policies to salespeople cross-sells: a cost-sensitive uplift modeling approach

Uplift modeling with quasi-loss-functions

Uplift modeling with quasi-loss-functions

Late Cretaceous Uplift of Grand Canyon: Evidence From Fluid Inclusions

For over a century, the history of Grand Canyon has been of interest to many. In recent years, debates have centered around the hypothesis that Grand Canyon formed during the late Cretaceous, not the Miocene, as previously thought. In this study, fluid inclusions within carbonates from the Mauv, Redwall, Supai, and Kaibab Fms. from Grand Canyon yield entrapment temperatures between 135 and 60 °C. Comparison of these temperature to time-temperature histories based on thermochronology (U-Th/He and fission track) from nearby samples suggest that these carbonates had fluids trapped within them from 89 to 58 Ma and that major denudation of late Cretaceous strata occurred during this interval. Regionally derived burial histories and local thermochronology suggest that significant uplift of Grand Canyon and the adjacent Colorado Plateau occurred during the late Cretaceous. We interpret the timing of fluid entrapment, denudation of Cretaceous strata, and burial histories to be consistent with initial uplift associated with the early stages of formation of Grand Canyon during the late Cretaceous. Models of uplift of northern Arizona exclusively during the Cenozoic are inconsistent with these data.

Shortening and asthenospheric flow drove Miocene vertical-axis rotations in the northern Tibetan Plateau: A palaeomagnetic study of the northeast Qaidam Basin of China

The timing and magnitude of upper crustal deformation in the northern Tibetan Plateau are important for testing models of plateau uplift and for understanding Cenozoic environmental variations and monsoon evolution in central and eastern Asia. Most published studies focus on the upward and outward growth of the northern Tibetan Plateau, whereas its tectonic, vertical-axis rotation history has received less attention. Here we evaluate paleomagnetic data from six stratigraphic sections in the northeastern Qaidam Basin of China, revealing a phase of clockwise rotation from ca. 20–11 Ma followed by counterclockwise rotation from ca. 11–5 Ma. We propose that differential crustal shortening and right-lateral shear drove early to middle Miocene clockwise rotation, whereas eastward flow of the Tibetan asthenosphere induced the post-middle Miocene counterclockwise rotation. These processes coincide with increased activity of first-order strike-slip faults in the northeastern plateau, emphasizing the important role of strike-slip faults in accommodating crustal deformation in northern Tibetan Plateau during the late Cenozoic.

Causal impact evaluation of occupational safety policies on firms’ default using machine learning uplift modelling

It is often undermined that occupational safety policies do not only displace a direct effect on work well-being, but also an indirect effect on firms’ economic performances. In such context, econometric models dominated the scenes of causality until recently while Machine Learning models were seen with skepticism. With the rise of complex datasets, an ever-increasing need for automated algorithms capable to handle complex non-linear relationships between variables has brought to uncover the power of Machine Learning for causality. In this paper, we carry out an evaluation of a public aid-scheme implemented in Italy and oriented to support investment of small and medium enterprises (SMEs) in occupational safety and health (OSH) for assessing the impact on the survival of corporations. A comparison of thirteen models is performed and the Individual Treatment Effect (ITE) estimated and validated based on the AUUC and Qini score for which best values of 0.064 and 0.407, respectively, are obtained based on the Light Gradient Boosting Machine (LightGBM). An additional in-depth statistical analysis also revealed that the best beneficiaries of the policy intervention are those firms that experience performance issues in the period just before the interventions and for which the increased liquidity brought by the policy may have prevented default.

Longitudinal analysis model for segment lining uplift during shield tunnelling considering shearing dislocation of circumferential joints

During shield tunnelling, longitudinal deformation occurs in the shield tunnel structure owing to the buoyancy generated by the synchronous slurry, which has adverse effects on the assembly quality and even the structural integrity of the segmental ring structure. However, the current longitudinal beam-like model for analysing the uplift response during shield tunnel construction adopts the Euler–Bernoulli beam to describe the shield tunnel. This model does not consider the shearing dislocation effect, causing errors in the structural longitudinal deformation analysis. In this context, this study proposes a shield tunnelling uplift model based on the Timoshenko beam theory, which can effectively consider the shearing dislocation effect, aiming to evaluate the longitudinal response of shield tunnels more accurately. A Pasternak two-parameter foundation with springs reflecting the resistance of the overburden with different thicknesses was adopted to simulate the supporting effects of the synchronous grouting layer and surrounding soil layer composite on the tunnel, considering the influences of both the shearing effect of the foundation and tunnel buried depth on the shield tunnel uplift. The dynamic buoyancy coefficient, determined through an inverse analysis of the field measurement data, was introduced to accurately calculate the magnitude of the dynamic buoyancy generated by the flow of synchronous slurry. The proposed model was validated by comparing its results with actual measurements. The uplift characteristics of the shield tunnel during construction were obtained by using the proposed model in an actual project. It was found that the longitudinal analysis model for shield tunnel uplift based on the Euler–Bernoulli beam theory underestimates the uplift and overestimates the internal forces generated during shield tunnelling.

Quantifying uncertainty of uplift: Trees and T-learners

Uplift modeling refers to the task of estimating the causal effect of a treatment on an individual, also known as the conditional average treatment effect. However, uplift models do not usually provide uncertainty estimates of the predictions. We explain why estimating uncertainty of the treatment effect is particularly important in many common use cases and we show how epistemic uncertainty of the uplift estimates can be quantified for T-learners and trees. We tested the methods on three empirical datasets and evaluated them on a simulated dataset. We found that high uncertainty might be the result of both modeling choices and properties of the data. Sometimes there is not enough data or the data is simply not rich enough to identify the treatment effect well resulting in high uncertainty. In addition, our results suggest that one commonly used dataset might not be suitable for benchmarking.

Propensity score oversampling and matching for uplift modeling

In this paper, we propose a novel matching strategy to correct for confounding in uplift modeling. Our method, called propensity score oversampling and matching (ProSOM), extends the well-known propensity score matching (PSM) technique by addressing one of its main limitations: dealing with small datasets that face an imbalance in the distribution of the causal variable. Apart from this, we also face the additional complexity of dealing with class labels. The proposed method establishes a parallel between uplift modeling and class-imbalance classification as it extends existing oversampling techniques to create synthetic elements from the treatment group. We design an algorithm that performs classaware data oversampling in the treatment group, and then it matches samples from this group with the control group. This can be seen as a novel hybrid undersampling-oversampling solution for causal learning. Experiments on five datasets show the virtues of ProSOM in terms of predictive performance, achieving the best Qini coefficient for all five datasets in relation to PSM and other resampling solutions.

Uplift modeling to predict individual treatment effects of renal replacement therapy in sepsis-associated acute kidney injury patients

Renal replacement therapy (RRT) is a crucial treatment for sepsis-associated acute kidney injury (S-AKI), but it is uncertain which S-AKI patients should receive immediate RRT. Identifying the characteristics of patients who may benefit the most from RRT is an important task. This retrospective study utilized a public database and enrolled S-AKI patients, who were divided into RRT and non-RRT groups. Uplift modeling was used to estimate the individual treatment effect (ITE) of RRT. The validity of different models was compared using a qini curve. After labeling the patients in the validation cohort, we characterized the patients who would benefit the most from RRT and created a nomogram. A total of 8289 patients were assessed, among whom 591 received RRT, and 7698 did not receive RRT. The RRT group had a higher severity of illness than the non-RRT group, with a Sequential Organ Failure Assessment (SOFA) score of 9 (IQR 6,11) vs. 5 (IQR 3,7). The 28-day mortality rate was higher in the RRT group than the non-RRT group (34.83% vs. 14.61%, p < 0.0001). Propensity score matching (PSM) was used to balance baseline characteristics, 458 RRT patients and an equal number of non-RRT patients were enrolled for further research. After PSM, 28-day mortality of RRT and non-RRT groups were 32.3% vs. 39.3%, P = 0.033. Using uplift modeling, we found that urine output, fluid input, mean blood pressure, body temperature, and lactate were the top 5 factors that had the most influence on RRT effect. The area under the uplift curve (AUUC) of the class transformation model was 0.068, the AUUC of SOFA was 0.018, and the AUUC of Kdigo-stage was 0.050. The class transformation model was more efficient in predicting individual treatment effect. A logistic regression model was developed, and a nomogram was drawn to predict whether an S-AKI patient can benefit from RRT. Six factors were taken into account (urine output, creatinine, lactate, white blood cell count, glucose, respiratory rate). Uplift modeling can better predict the ITE of RRT on S-AKI patients than conventional score systems such as Kdigo and SOFA. We also found that white blood cell count is related to the benefits of RRT, suggesting that changes in inflammation levels may be associated with the effects of RRT on S-AKI patients.

Uplift modeling to identify patients who require extensive catheter ablation procedures among patients with persistent atrial fibrillation

Identifying patients who would benefit from extensive catheter ablation along with pulmonary vein isolation (PVI) among those with persistent atrial fibrillation (AF) has been a subject of controversy. The objective of this study was to apply uplift modeling, a machine learning method for analyzing individual causal effect, to identify such patients in the EARNEST-PVI trial, a randomized trial in patients with persistent AF. We developed 16 uplift models using different machine learning algorithms, and determined that the best performing model was adaptive boosting using Qini coefficients. The optimal uplift score threshold was 0.0124. Among patients with an uplift score ≥ 0.0124, those who underwent extensive catheter ablation (PVI-plus) showed a significantly lower recurrence rate of AF compared to those who received only PVI (PVI-alone) (HR 0.40; 95% CI 0.19–0.84; P-value = 0.015). In contrast, among patients with an uplift score < 0.0124, recurrence of AF did not significantly differ between PVI-plus and PVI-alone (HR 1.17; 95% CI 0.57–2.39; P-value = 0.661). By employing uplift modeling, we could effectively identify a subset of patients with persistent AF who would benefit from PVI-plus. This model could be valuable in stratifying patients with persistent AF who need extensive catheter ablation before the procedure.

Calibration of Heterogeneous Treatment Effects in Randomized Experiments

Machine learning is commonly used to estimate the heterogeneous treatment effects (HTEs) in randomized experiments. Using large-scale randomized experiments on Facebook and Criteo platforms, we observe substantial discrepancies between machine learning-based treatment effect estimates and difference-in-means estimates directly from the randomized experiment. This paper provides a two-step framework for practitioners and researchers to diagnose and rectify this discrepancy. We first introduce a diagnostic tool to assess whether bias exists in the model-based estimates from machine learning. If bias exists, we then offer a model-agnostic method to calibrate any HTE estimates to known, unbiased, subgroup difference-in-means estimates, ensuring that the sign and magnitude of the subgroup estimates approximate the model-free benchmarks. This calibration method requires no additional data and can be scaled for large data sets. To highlight potential sources of bias, we theoretically show that this bias can result from regularization, and further use synthetic simulation to show biases result from misspecification and high-dimensional features. We demonstrate the efficacy of our calibration method using extensive synthetic simulations and two real-world randomized experiments. We further demonstrate the practical value of this calibration in three typical policy-making settings: a prescriptive, budget-constrained optimization framework; a setting seeking to maximize multiple performance indicators; and a multitreatment uplift modeling setting.

A Strategic Study on Improving Targeting Accuracy of Marketing Campaigns Using Factor Analysis

Abstract Precision marketing emerges as a pivotal mechanism to foster market development, enhancing the market economy and propelling internal economic growth. This study collates data pertinent to marketing activities, standardizes the raw datasets to compute descriptive statistics such as the mean and variance, and constructs a correlation coefficient matrix under specified conditions. Through factor analysis, the structure of this correlation matrix is meticulously examined. Factor loadings are employed to elucidate the relationships between factors and variables, thereby establishing a link between the precision of marketing endeavors and consumer attribute variables for further analytical probing. To assess the impact of implemented marketing strategies on consumer response behaviors, this research develops a precision marketing model using the Uplift algorithm. This model innovatively addresses the challenge of individual causal effects—which are realistically unresolvable—by transforming it into estimating the conditional average causal utility derivable from observed data. The factor analysis feasibility test yields a Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy at 0.792, and Bartlett's test of sphericity attains a significance level of 0.000, indicating robust factorability. Subsequent tests on variance and principal factor extraction reveal that five variables—such as income, gender, and age—are common factors across the datasets analyzed. Application of the Uplift model to the MineThatData and MegaFon datasets further substantiates the efficacy of the proposed marketing model. Particularly, the results from the MegaFon dataset validate the comprehensive applicability of this model, demonstrating its effectiveness in real-world scenarios.

Discrete Element Modelling of uplift of rigid pipes deeply buried in dense sand

This paper presents a numerical methodology based on the Discrete Element Method developed for the efficient modelling of kinematic granular soil-pipe interaction at large deformations. The methodology is based on a robust Bayesian procedure for calibrating micromechanical contact parameters using standard triaxial compression tests, does not rely on the back-analysis of physical model experiments, and produces accurate “blind” predictions of independent experimental measurements. The latter is demonstrated by extensively validating DEM models against measurements obtained from direct shear tests on sand performed during this study and published measurements from 1-g physical model experiments of uplift of rigid pipes buried in dense sand. In addition, we introduce different approaches that allow efficient modelling of deeply buried pipes, and we employ the methodology to investigate how the reaction from sand to rigid pipe uplift varies as the pipe embedment depth increases. Detailed numerical predictions provide insights into the “flow-around” failure mechanism that develops around the deeply buried pipes and on the existence of a critical embedment depth, beyond which the normalised reaction does not further increase with increasing pipe embedment. The outcomes of this study are applicable to the stress analysis of deeply buried pipes in practice and to the modelling of a variety of problems relevant to rigid objects buried deeply in granular soil.

Clumped Isotope Reordering and Kinetic Differences in Co-Hosted Calcite and Dolomite Minerals throughout Burial Diagenesis and Exhumation

The clumped isotope paleo-thermometer has become a valuable proxy for the burial history reconstruction of carbonate formations. To maximise the accuracy of these reconstructions, post-depositional alterations, such as recrystallisation and Δ47 isotope exchange reactions, must be understood. In this study, we examine the isotopic behaviour of calcites and early dolomite samples from the same stratigraphic intervals, and thus with similar burial history. This approach provides additional constraints on the kinetics of Δ47 reordering in dolomite during exhumation. Clumped isotope measurements were performed on 19 calcites and 15 early dolomites from the Permian, Jurassic, and Cretaceous periods from four locations in Oman spanning different burial regimes. The calcite and dolomite samples were collected from the rock matrix, based on the assumption that fine material was more susceptible to recrystallisation. Our results show that calcites and dolomites record different Δ47 values despite being subjected to the same thermal history. The maximum Δ47 temperature recorded in dolomites (181 ± 13 °C) corresponds to the oldest and most deeply buried Permian rock. This value is approximately 35 °C higher than those measured in the co-located and coeval calcite matrix (145 ± 14 °C). This discrepancy suggests that calcite and dolomite have different kinetic parameters. Our data confirm (1) that dolomite Δ47 values are more resistant to alteration during burial and exhumation than Δ47 calcite values, and (2) that dolomite has a higher Δ47 closing temperature than calcite during cooling. The presence of two mineral phases with distinct kinetic parameters in the same stratigraphic unit provides additional constraints on models of burial and uplift. In addition, mineralogical data coupled with Δ47 and burial depths suggest that the progressive development of dolomite cation ordering is driven by temperature elevation, as previously suggested.

Understanding the synergistic impact of stress release and cementation on sandstone using sound waves — Implications for exhumation estimation

Exhumation is the process that encompasses uplift and erosion, leading to the removal of overburden and the release of effective stress exerted on rocks. When estimating exhumation magnitude using the compaction trend method, it is commonly assumed that the physical properties of rocks are insensitive to stress reduction. However, recent laboratory evidence has indicated that porosity exhibits weaker sensitivity to stress release compared with velocity that can be significantly affected by stress release. This raises questions regarding the validity of irreversible compaction assumed by compaction trend method. It remains unclear whether the impact of stress release can be observed in real rocks in exhumed areas because there is a lack of methods to directly measure the impact of stress release on field data. In addition, studying real rocks is further complicated by the presence of rock diagenesis and its interaction with stress release. To address these knowledge gaps, this study uses stress-dependent burial and uplift modeling and interprets an extensive well-log data set using the modeling-derived evaluation metrics. Conceptual modeling suggests that methods that neglect the combined effect of cementation and stress release tend to underestimate the exhumation magnitude. Furthermore, we discover that the disparity between porosity sensitivity and velocity sensitivity to stress release can be leveraged to derive a metric we call “porosity inconsistency” that can serve as a qualitative and quantitative measure for identifying and evaluating stress release in sandstone using geophysical field measurements. We gather a significant amount of sonic velocity and porosity data from normally compacted and uplifted clean sandstones in the Norwegian Sea and the Barents Sea. Notably, we observe significant porosity inconsistency in the exhumed well 6510/2-1 in the Norwegian Sea. In the Barents Sea, which has experienced extensive Cenozoic exhumation, the well data reveal a varying pattern of porosity inconsistency increasing toward the north and decreasing toward the west. This distribution of porosity inconsistencies in the Barents Sea wells not only aligns with the spatial variation of exhumation reported in various studies but also exhibits a positive correlation with the magnitude of exhumation. Furthermore, the exhumation magnitude derived from velocity-depth trends is considerably lower than the magnitude obtained from porosity/density-depth trends for wells displaying significant porosity inconsistency. These observations provide support for the predictions made by the conceptual modeling. The results of this study enhance our understanding of the synergistic impact of stress release and cementation on sandstone. Moreover, these findings have implications for pore pressure prediction and core evaluation in exhumed areas. They also provide insights into the feasibility and interpretation of time-lapse data of reservoir injection, for which the effective stress is likely to decrease due to pore pressure buildup.

How to customize an early start preparatory course policy to improve student graduation success: an application of uplift modeling

How to customize an early start preparatory course policy to improve student graduation success: an application of uplift modeling