CONCEPTOS, MÉTODOS Y PRAXIS EN TÉCNICAS DE REGRESIÓN. APLICACIÓN A LAS CIENCIAS NATURALES

      

Luis M. Carrascal

Dept. Biogeografía y Cambio Global

Museo Nacional de Ciencias Naturales

Consejo Superior de Investigaciones Científicas

C/ José Gutiérrez Abascal, 2. 28006 Madrid. SPAIN

 

 

Literatura seleccionada:


* Hairston, N.G. (1989). Ecological Experiments: Purpose, Design and Execution. Cambridge Studies in Ecology, Cambridge Univ. Press, Cambridge.
* Scheiner, S.M.; Gurevitch, J. (2001). Design and analysis of ecological experiments. Chapman & Hall, New York.
* Keppel, G. (1991). Design and analysis: a researcher’s handbook. Prentice Hall, New Jersey. Nueva versión.
* Breiman, L. (1984). Classification and Regression Trees. Chapman & Hall
* Crawley, M.J. (1998). GLIM for Ecologists. Blackwell Science.
* Hastie, T.J.; Tibshirani, R.J. (1997). Generalized Additive Models. Chapman & may.
* Maxwell, S.E.; Delaney, H.D. (1990). Designing Experiments and Analyzing Data. A model comparison perspective. Wadsworth Publishing Company, Belmont, CA. Nueva versión.
* Burnham, K.P.; Anderson, D. (2003). Model Selection and Multi-Model Inference
Springer.
* Davison, A.C.; Hinkley, D.V. (2007). 
Bootstrap methods and their applicationCambridge Univ. Press. [link]

 

 

 

Programas:

PopTools – MS-Excell add-in para simulaciones

GRETL – para muchos modelos diferentes de regresión, incluyendo regresión de quantiles y series temporales

SAM – excelente paquete para estadística espacial y regresión usando Akaike – JAF Diniz Filho

TANAGRA – programa que interacciona con MS-Excel para “data mining” y regresión

G*Power – programa para análisis de potencia de los tests

R Project for Statistical Computing – entrono de trabajo R

RStudio – programa para uso de R

 

 

 

TEORÍA Y PRAXIS DE MODELOS GENERALIZADOS: INFIRIENDO PATRONES CON EL PAQUETE ESTADÍSTICO R. [link]

    Ejemplo de análisis con una variable respuesta con elevada carga de ceros: GLM y GAM.

    Ejemplo de análisis con variables predictoras muy relacionadas: PLS.

   

 

 

ASPECTOS GENERALES

   Examen de hipótesis, Hipótesis Nulas, Significación, Errores de tipo I y II, elección de tests. Potencia de los tests

Elementary Concepts in Statistics [pdf]

P values and statistical significance [pdf]

Errores de tipo I y II [pdf]

The logic of Hypotesis Testing
Null hypothesis testing
Citations Questioning the Indiscriminate Use of  Null Hypothesis Significance Tests in Observational Studies - Quotes Regarding Hypothesis Testing

Presentación general sobre estadística [pdf]

Trampas en los análisis de datos [pdf]
Selected reprints by David R. Anderson

Choosing statistical tests [pdf]

The Most Important Points

Potencia de los tests

Múltiples estimas de probabilidad

Interpreting Nonsignificant P values [pdf]

Resultados no significativos en investigación [pdf] [pdf] [pdf] [pdf] [pdf]

 

 

 

REGRESIÓN MÚLTIPLE (Modelos Generales Lineales)

   Referencias generales

Analysing ecological data – Linear regression

A Primer on Interpreting Regression Models

Linear Methods for Regression

Regression Methods in Biostatistics - Linear Regression

Simple means to improve the interpretability of regression coeficients

Multiple Regresión – NC State University

Presentación de conceptos (Wikipedia)
Regresión Múltiple - StatSoft

A protocol for data exploration to avoid common statistical problems

Limitations of Linear Regression Applied on Ecological Data

 

   Significación

When should we use one-tailed hypothesis testing?

Heads I win, tails you lose - testing directional alternative hypotheses in ecological and evolutionary research

Revised standards for statistical evidence

Ecologists should not use statistical significance tests to interpret simulation model results

   Diagnósticos de regresión

distancia de CookDFFITSresiduo studentizadoautocorrelación de residuos

multicolinearidadtoleranciafactor de inflado de la varianza (VIF)

 

   Transformaciones

Box-Cox Transformation: An Overview

The arcsine is asinine: the analysis of proportions in ecology

Spatial Autocorrelation: Trouble or New Paradigm? – sobre uso de términos polinomiales

 

   Residuos de modelos

On the misuse of residuals in ecology: regression of residuals vs. multiple regression
Problemas derivados del uso de residuos [
pdf] [pdf]

 

 

 

SIMULACIONES DE HIPÓTESIS NULAS, REMUESTREO, VALIDACIÓN

   Aspectos generales

What are cross-validation and bootstrapping? - Warren Sarle

Resampling, Bootstrap, Jackknife, Permutation tests – Bootstrap – Cross-validation - Monte Carlo method

Bootstrap Methods and their Application - Anthony Davison

Resampling Methods for statistical inference

Randomization, Bootstrap and Monte Carlo Methods in Biology

The Bootstrap, Permutation Tests, Simulation - Susan Holmes

Bootstrapping – David C. Howell

Null versus neutral models: what’s the difference? – N.J. Gotelli

   En regresión

Permutation tests for linear models

Resampling and Regression - Bob Andersen

Bootstrapping Regression Models - John Fox

Resampling techniques for statistical modeling - Gianluca Bontempi

 

 

 

PROCEDIMIENTOS DE SIMPLIFICACIÓN DE MODELOS

Reducción de modelos: Akaike information criterion [pdf] [pdf]; una síntesis de AIC

   Ejemplo de uso de AIC

Model Based Inference in the Life Sciences - A Primer on Evidence - Introduction

Model selection in ecology and evolution

AIC myths and misunderstandings

Avoiding Pitfalls When Using Information-Theoritic Methods

Information theory and hypothesis testing

Model selection and multimodel inference

Information theory and hypothesis testing - a call for pluralism

Concerns regarding a call for pluralism of information theory and hypothesis testing

Kullback-Leibler information as a basis for strong inference in ecological studies

Multimodel Inference: Understanding AIC and BIC in Model Selection

Multimodel inference

Model weights and the foundations of multimodel inference: Comment

Model del Inference and Averaging

Model Based Inference in the Life Sciences - A Primer on Evidence - Quantifying the Evidence About Science Hypotheses

Model Based Inference in the Life Sciences - A Primer on Evidence – Appendices … interesantes los E y F

Statistical Modeling by AIC

Bootstrap Information Criterion

Information criterion

Information theory in willdlife science - critique and viewpoint

Uninformative parameters and model selection using Akaike’s Information Criterion

Testing ecological theory using the information-theoretic approach - Examples and cautionary results

Why do we still use stepwise modelling in ecology and behaviour?

Performance of several variable-selection methods applied to real ecological data

Reducción de modelos: Mallows' Cp, [link]

… hoja de MS-Excel para calcular AIC en modelos GLM – trabajando con salidas de STATISTICA

 

 

 

MODELOS LINEALES GENERALIZADOS (regresión múltiple)

Generalized Linear Models

Generalised Linear Modelling

Concepto de maximum likelihood

Síntesis y presentación general

Generalized Linear Models – Simon Jackman

The negative binomial model

Logistic Regression

Logistic Regression – NC State University

Generalized linear and generalized additive models in studies of species distributions: setting the scene – A. Guisan

Generalized linear mixed models: a practical guide for ecology and evolution

Generalized Linear Models (GLZ) – StatSoft

Receiver Operating Characteristic (ROC) [excelente link] [otros vínculos]

 

 

 

MODELOS GENERALIZADOS ADITIVOS (regresión múltiple)

Additive and generalised additive modelling

GLM and GAM for Count Data

Regression Splines and Regression Smoothers

Spline Smoothing

Local Smoothers

Generalized Additive Models (GAM) - StatSoft

Additive Models, Trees, and Related Methods

 

 

 

REGRESIÓN POR EL MÉTODO DE LOS MÍNIMOS CUADRADOS PARCIALES (PLS)

A Beginner’s Guide to Partial Least Squares Analysis

Partial Least Squares (PLS) Regression. - H. Abdi – otra versión

An Introduction to Partial Least Squares Regression - Randall D. Tobias
An lnterpretation of Partial Least Squares - Paul H. Garthwaite

Partial least squares regression as an alternative to most currently used regression methods in Ecology

PARTIAL LEAST SQUARES REGRESSION

Partial Least Squares - StatSoft

... cómo usarlo con Statistica y Tanagra

 

 

REGRESIÓN DE QUANTILES

A gentle introduction to quantile regression for ecologists

Quantile regression reveals hidden bias and uncertainty in habitat models

An introduction to quantile regression

Quantile Regression

Quantile Regression - Marzban

Tutorial on quantile regression

Nonparametric Quantile Regression

Computational Issues for Quantile Regression

Quantile regression - applications and current research areas

 

 

 

ARBOLES DE REGRESIÓN Y CLASIFICACIÓN

Árboles únicos

Recursive Partitioning and Tree-Based Methods

Classification and Regression Trees (CART)

Classification and regression trees: a powerful yet simple technique for ecological data analysis - G. DE’ATH & K. E. FABRICIUS

Classification And Regression Trees: An Introduction – Y. Yohannes & J. Hoddinott

Comparación de los Árboles de Clasificación con otras técnicas de regresión

An evaluation of methods for modelling species distributions


Aproximaciones ‘bagging’ y ‘boosted’
A working guide to boosted regression trees
Boosted trees for ecological modeling and prediction

Boosting and Additive Trees
Modern Multivariate Statistical Techniques - Committee Machines

Newer Classification and Regression Tree Techniques - Bagging and Random Forests for Ecological Prediction
Boosted regression (boosting)

Bagged Averaging of Regression Models

Obtaining Calibrated Probabilities from Boosting

... cómo usarlo con Statistica 

 

Actualización: 10/09/2015