میانی
تحلیل رگرسیون
کمترین مربعات و مفروضات آن، رگرسیون چندگانه و تشخیصهای آن، دادههای گمشده، رگرسیون لجستیک، کاپلان-مایر و مدل کاکس، بررسی مخاطرههای متناسب، مدلهای چندسطحی و طولی، و گزارش صادقانه نتیجه یک رگرسیون.
فهرست درسها
01Correlation and Simple Linear RegressionMeasuring the strength of a linear association and then specifying it. Why a scatter plot must come first, how correlation and slope are related, and why the regression of Y on X differs from the regression of X on Y.انگلیسی45 دقیقه02Assumptions and Least Squares EstimationWhere regression came from, what the model actually assumes, how least squares picks the line, and the properties that make the estimates trustworthy. Includes regression to the mean, the phenomenon that gave the method its name.انگلیسی45 دقیقه03Inference in Regression and the Move to Multiple PredictorsThe ANOVA table for a regression, tests and intervals for the intercept and slope, and the extension to more than one predictor, where coefficients become adjusted effects and start depending on what else is in the model.انگلیسی45 دقیقه04Multiple Regression, Categorical Predictors, and DiagnosticsComparing nested models with the partial F-test, coding categorical predictors with dummy variables, using residuals to find outliers and influential points, and transforming variables when the assumptions fail.انگلیسی50 دقیقه05Missing DataWhy missing data matters, the three mechanisms and why two of them are indistinguishable, the deletion and single-imputation methods and what each destroys, and multiple imputation done properly.انگلیسی40 دقیقه06Logistic RegressionModeling a binary outcome. Maximum likelihood instead of least squares, deviance instead of residual sum of squares, likelihood ratio tests instead of partial F-tests, and AIC and BIC for comparing models that are not nested.انگلیسی45 دقیقه07Logistic Regression Diagnostics and ExtensionsDetecting collinearity and influential observations when there is no residual sum of squares to work with, sample size for logistic models, and the variants for matched designs and outcomes with more than two levels.انگلیسی45 دقیقه08Survival Analysis FoundationsWhy time-to-event data needs its own methods, the three functions that describe it, the assumptions censoring must satisfy, and the Kaplan-Meier estimator and log-rank test.انگلیسی45 دقیقه09Cox Proportional Hazards RegressionA regression model for censored outcomes that leaves the baseline hazard unspecified. The partial likelihood, why it only uses the order of events, how ties are handled, and how to interpret a hazard ratio.انگلیسی45 دقیقه10Checking Proportional Hazards, and Parametric ModelsGraphical and formal ways to detect violations of proportional hazards, residuals for the Cox model, what to do when the assumption fails, and the parametric alternatives that model the baseline hazard directly.انگلیسی45 دقیقه11Multilevel and Longitudinal ModelsWhat to do when the independence assumption fails. Clustered and repeated-measures data, random intercepts and random slopes, the intraclass correlation, and the choice between mixed models and GEE.انگلیسی45 دقیقه12Reporting Statistical AnalysesHow to report an analysis so that it can be understood, checked, and pooled. Common errors, the statistical terms that are routinely misused, conventions for numbers and decimal places, and what belongs in tables and figures.انگلیسی40 دقیقه