Preguntas etiquetadas con hypothesis-testing

8
¿Por qué un modelo estadístico se sobreajusta si se le da un gran conjunto de datos?

Mi proyecto actual puede requerir que construya un modelo para predecir el comportamiento de un determinado grupo de personas. el conjunto de datos de entrenamiento contiene solo 6 variables (la identificación es solo para fines de identificación): id, age, income, gender, job category, monthly...

modeling large-data overfitting clustering algorithms error spatial r regression predictive-models linear-model average measurement-error weighted-mean error-propagation python standard-error weighted-regression hypothesis-testing time-series machine-learning self-study arima regression correlation anova statistical-significance excel r regression distributions statistical-significance contingency-tables regression optimization measurement-error loss-functions image-processing java panel-data probability conditional-probability r lme4-nlme model-comparison time-series probability probability conditional-probability logistic multiple-regression model-selection r regression model-based-clustering svm feature-selection feature-construction time-series forecasting stationarity r distributions bootstrap r distributions estimation maximum-likelihood garch references probability conditional-probability regression logistic regression-coefficients model-comparison confidence-interval r regression r generalized-linear-model outliers robust regression classification categorical-data r association-rules machine-learning distributions posterior likelihood r hypothesis-testing normality-assumption missing-data convergence expectation-maximization regression self-study categorical-data regression simulation regression self-study self-study gamma-distribution modeling microarray synthetic-data
8
¿Se puede usar BIC para pruebas de hipótesis?

Defina el criterio de información bayesiano como (no dejo caer el constante, , para evitar problemas cuando se compara con la probabilidad marginal)BIC=−2⋅lnL^+k⋅(ln(n)−ln(2π))BIC=−2⋅ln⁡L^+k⋅(ln⁡(n)−ln⁡(2π)) \mathrm{BIC} = {-2 \cdot \ln{\hat L} + k \cdot (\ln(n) - \ln(2 \pi))} −ln(2π)−ln⁡(2π) -...

hypothesis-testing bayesian bic
8
Cómo realizar SVD para imputar valores perdidos, un ejemplo concreto

He leído los excelentes comentarios sobre cómo lidiar con los valores perdidos antes de aplicar SVD, pero me gustaría saber cómo funciona con un ejemplo simple: Movie1 Movie2 Movie3 User1 5 4 User2 2 5 5 User3 3 4 User4 1 5 User5 5 1 5 Dada la matriz anterior, si elimino los valores de NA,...

r missing-data data-imputation svd sampling matlab mcmc importance-sampling predictive-models prediction algorithms graphical-model graph-theory r regression regression-coefficients r-squared r regression modeling confounding residuals fitting glmm zero-inflation overdispersion optimization curve-fitting regression time-series order-statistics bayesian prior uninformative-prior probability discrete-data kolmogorov-smirnov r data-visualization histogram dimensionality-reduction classification clustering accuracy semi-supervised labeling state-space-models t-test biostatistics paired-comparisons paired-data bioinformatics regression logistic multiple-regression mixed-model random-effects-model neural-networks error-propagation numerical-integration time-series missing-data data-imputation probability self-study combinatorics survival cox-model statistical-significance wilcoxon-mann-whitney hypothesis-testing distributions normal-distribution variance t-distribution probability simulation random-walk diffusion hypothesis-testing z-test hypothesis-testing data-transformation lognormal r regression agreement-statistics classification svm mixed-model non-independent observational-study goodness-of-fit residuals confirmatory-factor neural-networks deep-learning