Articles
Permanent URI for this collectionhttp://repository.kln.ac.lk/handle/123456789/15894
Browse
4 results
Search Results
Item Temporal cross-validation in forecasting: A case study of COVID-19 incidence using wastewater data(Quality and Reliability Engineering International, 2024-11) Lai, M.; Wulff, S. S.; Cao, Y.; Robinson, T. J.; Rajapaksha, R.Two predominant methodologies in forecasting temporal processes include traditional time series models and machine learning methods. This paper investigates the impact of time series cross-validation (TSCV) on both approaches in the context of a case study predicting the incidence of COVID-19 based on wastewater data. The TSCV framework outlined in the paper begins by engineering interpretable features hypothesized as potential predictors of COVID-19 incidence. Feature selection and hyperparameter tuning are then utilized with TSCV to identify the best features and hyperparameters for optimal model performance given a specific forecast horizon. While evidence supporting the utility of TSCV for auto-regressive integrated moving average model with exogenous variables (TS-ARIMAX) forecasts is lacking in this study, such an approach proves advantageous for gradient boosting machine forecasts (TS-GBM). In Wyoming, for instance, TS-GBM had a 34.9% improvement compared to naïve predictions, whereas GBM without TSCV only had a 15.6% improvement. However, TSCV also enhances interpretability for both TS-ARIMAX and TS-GBM models as this approach selects specific features, such as lagged values of COVID-19 cases, based on forecast performance and forecast length. Future research should work to explore the influence of stationarity and model averaging on the performance of TSCV in forecasting applications.Item Proposed hybrid approach for three-dimensional subsurface simulation to improve boundary determination and design of optimum site investigation plan for pile foundations(Soils and Foundations, 2023) Oluwatuyi, O. E.; Rajapakshage, R.; Wulff, S. S.; Ng, K.Geological uncertainty refers to the changeability of a geomaterial category embedded in another. It arises from predicting a geomaterial category at unobserved locations using categorical data from a site investigation (SI). In the design of bridge foundations, geological uncertainty is often not considered because of the difficulties of assessing it using sparse borehole data, validating the quality of predictions, and incorporating such uncertainties into pile foundation design. To overcome these problems, this study utilizes sparse borehole data and proposes a hybrid approach of various spatial Markov Chain (spMC) models and Monte Carlo simulation to predict three-dimensional (3D) geomaterial categories and assess geological uncertainties. The 3D analysis gives realistic and comprehensive information about the site. Characteristics of the proposed hybrid approach include the estimation of transition rates, prediction of 3D geomaterial categories, and simulation of multiple realizations to propagate the uncertainties quantified by information entropy. This proposed hybrid approach leads to specific novelties that include the development of optimal SI plans to reduce geological uncertainty and the determination of geomaterial layer boundaries according to the quantified geological uncertainty. Reducing the geological uncertainties and accurately determining spatial geomaterial boundaries will improve the design reliability and safety of bridge foundations. The hybrid approach is applied to the Lodgepole Creek Bridge project site in Wyoming to demonstrate the application of the hybrid approach and the associated novelties. Outcomes are cross-validated to evaluate the geomaterial prediction accuracy of the hybrid approach.Item Optimal Site Investigation Through Combined Geological and Property Uncertainties Analysis(Geotechnical and Geological Engineering, 2023) Oluwatuyi, O. E.; Ng, K.; Wulff, S. S.; Rajapakshage, R.Site investigation is crucial in character- izing the geomaterial profile for the design of bridge pile foundations. A site investigation plan should be conducted to maximize geomaterial information and minimize uncertainty. Thus, both geological and property uncertainties should be explicitly incorpo- rated into a site investigation plan. This leads to the question of how to choose the corresponding optimal number and location of boreholes in a multiphase site investigation plan in order to reduce these uncer- tainties. This study addresses these problems using multinomial categorical prediction and universal kriging on a random field with multiple simulations. Site investigation data for this study are taken from a bridge project in Iowa, USA, which consists of four boreholes, each within the proximity of the pile foundation location. Subsequent numbers of recom- mended boreholes and their associated locations are determined to minimize the combined uncertain- ties. The effectiveness of this combined analysis for determining an optimal site investigation plan (OSIP) is validated and compared to an analysis done solely on property uncertainty. The proposed OSIP yields a lower prediction error, improves the prediction of geomaterial type and property, and reduces the sub- surface uncertainties. The incorporation of OSIP invariably improves the design efficiency and perfor- mance of bridge pile foundationsItem Inherent variability assessment from sparse property data of overburden soils and intermediate geomaterials using random field approaches(Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards, 2022) Oluwatuyi, O. E.; Holt, R.; Rajapakshage, R.; Wulff, S. S.; Ng, K.This study assesses the inherent variability in the geomaterial parameter by quantifying the parameter uncertainty and develops a site investigation plan with a low degree of uncertainty. A key research motivation was using sparse borehole data to predict a site geomaterial configuration in order to determine the design of a site investigation plan. This study develops a systematic methodology for carrying out a study of inherent variability in light of the limitations posed by borehole data. The data in this study was provided by the Iowa Department of Transportation which consisted of eight boreholes from which 92 associated SPT N-values was considered as the geomaterial parameter of interest. The systematic methodology then involved the following steps. A general linear model was employed to fit and compare various spatial covariance models with and without a nugget. These spatial covariance models were also evaluated with variograms. Predicted SPT N-values were generated using universal kriging. Simulations were performed conditionally and unconditionally to identify optimal site investigation plans. The results identified site investigation plans with reduced parameter uncertainty. The proposed approach can produce site investigation plans that target any or all geomaterial layers to reduce uncertainty with respect to any geomaterial parameter of interest.