UID:
edoccha_9961612705302883
Umfang:
1 online resource (532 pages)
Ausgabe:
1st ed.
ISBN:
9783031609466
Serie:
Water Science and Technology Library ; v.108
Anmerkung:
Intro -- Contents -- About the Author -- 1 Introduction to Missing Data -- 1.1 Introduction -- 1.2 Spatial and Temporal Data -- 1.2.1 Working with Gridded and Point Data -- 1.2.2 Gridded Data: Bias Corrections -- 1.2.3 Scale Triplet -- 1.3 Missing Data: Causes -- 1.3.1 Nature of Errors: Precipitation Measurements -- 1.3.2 Addressing Missing Data -- 1.4 Mechanisms of Missing Data -- 1.5 Missing Data Patterns -- 1.5.1 Implications of Ignoring Missing Data -- 1.5.2 Documenting Missing and Estimated Records -- 1.6 Handling Missing Data: Need for Imputation Methods -- 1.7 Missing Data Estimation Methods -- 1.8 Summary -- References -- 2 Imputation Methods: An Overview -- 2.1 Introduction: Imputation Methods -- 2.2 Point Estimation and Surface Generation -- 2.3 Local and Global Interpolation -- 2.4 Exact and Inexact Interpolation -- 2.5 Missing Data Tolerance Methods -- 2.5.1 List-Wise Deletion/Complete-Case Analysis -- 2.6 Naïve Methods -- 2.6.1 Hot Deck Imputation Method -- 2.7 Requirements of Spatial and Temporal Interpolation Methods -- 2.8 Geostatistical Methods -- 2.9 Trend Surface Methods -- 2.10 Data-Driven Methods -- 2.10.1 Methods Using Data Mining Concepts -- 2.10.2 Machine Learning (ML) Methods -- 2.11 Single and Multiple Imputation -- 2.12 Statistics Preserving Methods -- 2.13 Simple Corrections of Spatially Interpolated Estimates -- 2.14 Summary -- References -- 3 Temporal Interpolation Methods -- 3.1 Introduction -- 3.2 Autocorrelation -- 3.3 Conceptually Simple Imputation Methods -- 3.3.1 Nearest-Neighbor Imputation -- 3.3.2 Mean Imputation Method -- 3.3.3 Random Value Replacement Method -- 3.3.4 Seasonally Decomposed Imputation -- 3.4 Function Approximation-Based Interpolation Methods -- 3.4.1 Linear Models -- 3.4.2 Nonlinear Models -- 3.5 Sequential Temporal Imputation Methods.
,
3.5.1 Baseline Observation Carried Forward (BOCF) Method -- 3.5.2 Last Observation Carried Forward (LOCF) Method -- 3.6 Imputation with Time Series-Based Models -- 3.6.1 Autoregressive and Moving Average Models -- 3.7 Smoothing Methods -- 3.8 Universal Function Approximation Method -- 3.9 Methods with Exogenous Variables -- 3.10 Interpolation in Extrapolation Mode -- 3.11 Multiple Imputation -- 3.12 Forward and Backward Forecasting-Based Imputation -- 3.13 Regularized Regression Models -- 3.13.1 Regularization Process -- 3.13.2 Ridge Regression -- 3.13.3 LASSO Regression -- 3.13.4 Geometric Interpretation of L2 and L1 Regularizations -- 3.13.5 Elastic Net -- 3.13.6 Development of Regularized Regression Models -- 3.13.7 Illustrative Example of Regularized Regression -- 3.13.8 Fine Tuning Regularization Parameters -- 3.13.9 Advantages and Limitations of Regularization Methods -- 3.14 Objective Selection of Predictor Variables -- 3.15 Kernel Regression -- 3.15.1 GKR for Discharge Data -- 3.15.2 GKR for Temperature Data -- 3.16 Utility of Temporal Interpolation -- 3.17 Computational Packages and Software for Univariate Imputation -- 3.18 Limitations of Temporal Interpolation Methods -- 3.19 Summary -- References -- 4 Spatial Interpolation Methods -- 4.1 Introduction -- 4.1.1 Definitions -- 4.1.2 Notation -- 4.2 Naive Approaches -- 4.2.1 Single Best Estimator (SBE) -- 4.2.2 Site Mean Estimator (SME) -- 4.2.3 Climatological Mean Estimator (CME) -- 4.3 Weighting Methods -- 4.3.1 Euclidean Distance-Based Methods -- 4.3.2 Illustrative Example of IDWM Application -- 4.4 Variants of IDWM -- 4.4.1 Integration of Thiessen Polygon Approach and Inverse Distance Method -- 4.4.2 Inverse Exponential Weighting Method (IEWM) -- 4.5 Quadrant Method -- 4.6 Coefficient of Correlation Weighting Method (CCWM) -- 4.7 Surrogates for Euclidean Distances in Weighting Methods.
,
4.8 Weights-Based on Proximity Metrics -- 4.8.1 Distance Metrics for Precipitation Data -- 4.9 Binary Distance Measures -- 4.10 Optimization Approaches: Regression -- 4.10.1 Regression Models -- 4.10.2 Multiple Linear Regression (MLR) -- 4.10.3 Non-negative Least-Squares (NLS) Method -- 4.10.4 Stochastic Regression -- 4.11 Mathematical Programming Methods -- 4.11.1 Optimum Weighting Models -- 4.11.2 Model I -- 4.11.3 Model IA -- 4.11.4 Model II -- 4.11.5 Model IIA -- 4.11.6 Model IIB: Stratification of Data -- 4.11.7 Model III -- 4.11.8 Model IIIA -- 4.11.9 Model IV: Site Clusters -- 4.11.10 Model VA -- 4.11.11 Model VB -- 4.11.12 Optimal Gauge Mean (GME) and Single Best Estimator (SBE) -- 4.11.13 Single or Multiple Best Estimator-Based Corrections -- 4.11.14 Correlation Weighting Approach Using Best Estimators -- 4.11.15 Goal Programming Formulation -- 4.11.16 Issues with Objective Functions -- 4.12 Shepard's Method -- 4.12.1 Neighborhood Size Selection -- 4.12.2 Association Measure: Correlation Coefficient -- 4.12.3 Distribution Similarity -- 4.12.4 Correlation and Distributional Similarity -- 4.12.5 Methodology -- 4.13 Geostatistical Method: Kriging -- 4.13.1 Modeling Spatial Variability: Variogram -- 4.13.2 Estimation of Missing Data: Illustration of Ordinary Kriging (OK) Application -- 4.13.3 Positive Kriging -- 4.13.4 Authorized Semi-variogram Models -- 4.13.5 Isotropy and Neighborhood -- 4.13.6 Considerations for Development of Variograms -- 4.13.7 Precipitation-Specific Kriging Applications -- 4.13.8 Limitations of Kriging -- 4.14 Surface Fitting Methods -- 4.14.1 Examples of Trend Surface Models -- 4.14.2 Validity of Trend Surface Model: Use of Analysis of Variance -- 4.14.3 Thin Plate Splines -- 4.14.4 Splines with Tension -- 4.14.5 Multiquadric Surface Method -- 4.14.6 Surface Fitting Methods: Illustrative Example.
,
4.14.7 Surface Generation Methods: Issues -- 4.15 Natural Neighbor Method -- 4.16 Methods Using Auxiliary Information -- 4.17 Methods for Estimation of Missing Gridded Data -- 4.17.1 Estimation Using Nearby Gauge or Sensor -- 4.18 Missing Data Estimation for Gridded Data -- 4.19 Geo-spatial Grid-Based Interpolation: Example Using Radar-Based Precipitation Data -- 4.19.1 Area Weighting Method -- 4.19.2 Maximum Area Method -- 4.19.3 Inverse Distance Weighting Method -- 4.19.4 Equal Weights (Average) Method -- 4.20 Disaggregation of Gridded or Point Data -- 4.20.1 Method of Fragments -- 4.20.2 Single Best Neighbor (SBN) -- 4.20.3 Function Approximation Methods -- 4.20.4 Methods Using Multiple Neighbors -- 4.20.5 Optimization Methods -- 4.20.6 Pattern-Based Methods -- 4.21 Statistics Preserving Spatial Interpolation -- 4.21.1 Optimization Formulations -- 4.22 Corrections of Interpolated Estimates -- 4.22.1 Quantile Mapping (QM) -- 4.22.2 Equi-ratio Quantile Matching (ERQM) -- 4.23 Imputation Using Estimates -- 4.23.1 Rain Gauge-Radar Functional Relationships -- 4.24 Geographically Weighted Optimization -- 4.25 Computationally Intensive Estimation Methods -- 4.26 Optimization Issues: Solvers and Solution Methods -- 4.27 Data Filler Approaches: Application in Real-Time -- 4.28 Spatial Interpolation: Issues -- 4.29 Under- and Overestimations -- 4.30 Monitoring Networks -- 4.31 Software and Spatial Analysis Environments for Interpolation -- 4.32 Use of Spatial Interpolation Methods: Considerations -- 4.33 Summary -- References -- 5 Universal Function Approximation and Data Mining-Assisted Imputation Methods -- 5.1 Introduction: Optimal Functional Forms and Function Approximation -- 5.2 Fixed Function Relationships -- 5.2.1 Fixed Function Set Genetic Algorithm Method FFSGAM -- 5.2.2 FFSGAM for Estimating Missing Precipitation Data.
,
5.2.3 Mathematical Programming Formulation for Optimal Coefficients -- 5.2.4 Application of FFSGAM: Issues -- 5.3 Data Mining Methods -- 5.3.1 Corrections of Interpolated Estimates: Use of Data Mining Methods -- 5.3.2 Association Rule Mining (ARM) -- 5.3.3 Support -- 5.3.4 Confidence -- 5.3.5 Association Rule Mining-Supported Spatial Interpolation -- 5.3.6 Data Mining Tool: WEKA -- 5.4 Universal Function Approximators: Artificial Neural Networks (ANNs) -- 5.4.1 Architecture of ANN -- 5.4.2 Activation Functions -- 5.4.3 Calculation of Outputs -- 5.4.4 Building ANN Using Optimization Formulation -- 5.4.5 Radial Basis Function Neural Network (RBFNN) -- 5.4.6 ANN Training and Function Approximation Capabilities -- 5.4.7 Preparing Data for ANNs -- 5.4.8 Imputation of Missing Data Using ANN -- 5.4.9 ANNs for Temporal Data Disaggregation -- 5.4.10 Elicitation of Knowledge from ANNs -- 5.5 Universal Function Approximation-Based Kriging -- 5.6 ANNs for Imputation: Issues -- 5.7 Computational Packages and Software -- 5.8 Summary -- References -- 6 Machine Learning and Multiple Imputation Methods -- 6.1 Introduction -- 6.1.1 Definitions -- 6.2 Development of Machine Learning-Based Models -- 6.3 Imputation of Missing Data and ML -- 6.4 Decision and Regression Trees -- 6.4.1 Decision Tree -- 6.4.2 Regression Tree -- 6.5 Regression Trees and Imputation Problem -- 6.5.1 Variants of Regressions Tree -- 6.5.2 Illustrative Example Using Regression Tree -- 6.5.3 Results and Analysis -- 6.6 Model Tree Approach -- 6.6.1 Illustrative Example Using Model Tree Approach -- 6.7 Advantages and Limitations of Trees -- 6.8 Ensemble Modeling Approaches -- 6.9 Multiple Imputation Methods -- 6.9.1 Multiple Imputations -- 6.10 Bagging Method -- 6.10.1 Boosting Method -- 6.11 Random Forest Method -- 6.12 Illustrative Example Using Bagging and RF Methods.
,
6.13 Ensemble Learning Methods: Summary.
Weitere Ausg.:
Print version: Teegavarapu, Ramesh S. V. Imputation Methods for Missing Hydrometeorological Data Estimation Cham : Springer International Publishing AG,c2024 ISBN 9783031609459
Sprache:
Englisch
Bookmarklink
