# format_iboost: Function to calculate selective inference based on sampled... In davidruegamer/iboost: Inference for model-based boosting

## Description

This function calculates p-values and confidence intervals based on sampled values and weights obtained by the `iboost` function.

## Usage

 `1` ```format_iboost_res(res, alpha, method, vT, this_var, computeCI, fac = 2, y) ```

## Arguments

 `res` list of sampled values either directly obtained in the `iboost` call or when using `iboost` with `returnSamples = TRUE`. `alpha` numeric; level for confidence interval(s) `method` character; method used for sampling `vT` list of test vector(s) `this_var` numeric; variance to be used for inference `computeCI` logical; whether or not to compute confidence intervals `fac` numeric; used in inverse search for confidence interval limits `y` vector; response vector

## Details

The specific use of this function is to (internally) compute inference based on samples in the `iboost` function, to recalculate inference for given samples `dist` from an `iboost` object or to calculate inference in the first place when `iboost` has been ran with argument `returnSamples = TRUE`. In the first and third case, use `format_iboost_res`, otherwise `format_iboost`.

## Examples

 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32``` ```if(require("mboost")){ set.seed(0) n <- 200 x1 <- rnorm(n) x2 <- rnorm(n) + 0.25 * x1 x3 <- rnorm(n) eta <- 3 * sin(x1) + x2^2 y <- scale(eta + rnorm(n), scale = FALSE) spline1 <- bbs(x1, knots = 20, df = 4) knots.x2 <- quantile(x2, c(0.25, 0.5, 0.75)) spline2 <- bbs(x2, knots = knots.x2, df = 4) spline3 <- bbs(x3, knots = 20, df = 4) data <- data.frame(y=y, x1=x1, x2=x2, x3=x3) mod1 <- mboost(y ~ spline1 + spline2 + spline3, control=boost_control(mstop = 73), offset = 0, data = data) # calculate p-values and intervals for model with # fixed stopping iteration: # this is done with only B = 100 samples for # demonstrative purposes and should be increased # for actual research questions res <- iboost(mod1, method = "impsamp", B = 100) # recalculate inference for different variance or alpha level format_iboost(res, alpha = 0.1, this_var = var(y)*5) } ```

davidruegamer/iboost documentation built on May 14, 2019, 3:10 a.m.