lasso: Use a LASSO algorithm to find important predictors

In haranse/dyadmlvar: Dyadic Timeseries Extension for Mlvar

Description

lasso Function that uses the LASSO shrinkage reduction method to find meaningful predictors of an outcome vector

Usage

lasso(Data, Predictors, Outcome, Seeds = 1, Train = F,
  PropOfTrain = 0.75, Plot = F)

Arguments

Data	a dataframe with columns for each predictor and for the outcome variable
Predictors	a list of strings with predictor variables names
Outcome	a string with the name of an outcome variable
Seeds	a seed for randomly selecting some part of the data to be training data and some to be testing data. default -
Train	Should we split data to training and test datasets (FALSE uses all data for both)
PropOfTrain	How much of the data to use for training the model
Plot	shoud we show a plot of the parameter number and log likelihood?

Value

a list of estimates for the effect of the valid predictors, and a R^2 statistic for the final model

Examples

ntwrk <- read_network(fit1)
#intra-partner variables from the partner B
intra_B_vars <- get_names(ntwrk, part = INTRA_B, time = ALL_NETWORK)
full_data <- merge(ntwrk$all, sat, by = "ID", all=TRUE, sort=TRUE)
lasso(full_data, intra_B_vars, "W_csi_resid")

haranse/dyadmlvar documentation built on May 23, 2019, 6:06 p.m.