make_possible_data_single: Make possible data for a single strategy step

In lilymedina/gbiqqtools: Tools for generalized BIQQ

Description

Creates a database of possible data from a data strategy. Users can gather additional data on variables specified via vars for any possible cases in the model ("any"). Or they can gather data in all cases within a given dataset ("within"). Or they can specify the subset of cases for which within-case data should be collected (e.g. "Y == 1").

Usage

make_possible_data_single(model, given = NULL, N = 1, within = FALSE,
  condition = TRUE, vars = NULL)

Arguments

model	A causal model as created by make_model
given	A data frame in compact form with first column indicating event type and second column indicating number of events of that type.
N	Number of variables to seek
within	logical Whether to seek variables within existing data
condition	A list of character strings indicating for which cases data should be gathered. Options are: (i) to gather additional data on variables specified via vars for any possible cases in the model ("any"), (ii) to gather data in all cases within a given dataset ("within"), or (iii) to specify the subset of cases for which within-case data should be collected (e.g. "Y == 1").
vars	Variables to be sought or NA. If NA make_possible_data gathers data on all variables containing NA for the specified data strategy.

Value

A dataset

Examples

library(dplyr)
model <- make_model("X->M->Y")  %>%
   set_restrictions(causal_type_restrict = "Y[M=1]<Y[M=0] | M[X=1]<M[X=0] ") %>%
   set_parameter_matrix()
df <- data.frame(X = c(0,0,0,1,1,1), M = NA, Y = c(0,0,1,0,1,1))
given <- summarize_data(model, df)[, -2]

# Look for data on M for all possible cases in the given data
make_possible_data_single(model, N = 2)
make_possible_data_single(model, given = given, vars = "M", within = TRUE, N = 2)
make_possible_data_single(model, given = given,
                          within = TRUE, vars = "M",
                          N = 2,
                          condition = "X==1 & Y==1")

model <- make_model("X -> M -> Y <- K")  %>%
   set_restrictions(causal_type_restrict = "(Y[M=1, K= .]<Y[M=0, K= .]) | M[X=1]<M[X=0] ") %>%
   set_parameter_matrix()
given <- data.frame(X = c(0,0,0,1,1,1), K = NA,  M = NA, Y = c(0,0,1,0,1,1)) %>%
         collapse_data(model = model)
make_possible_data_single(model, given = given,
                          within = TRUE,
                          N = 2,
                          condition = "X==1",
                          vars = "M")

make_possible_data_single(model, given = given,
                          within = TRUE,
                          N = 2,
                          condition = "X==1 & Y==1",
                          vars = c("M", "K"))

model <- make_model("X->M->Y")  %>%
set_restrictions(causal_type_restrict = "Y[M=1]<Y[M=0] | M[X=1]<M[X=0]") %>%
set_parameter_matrix()
given <- data.frame(X = c(0,0,0,1,1,1), M = NA, Y = c(0,0,1,0,1,1)) %>%
         collapse_data(model)
make_possible_data_single(model,
                          given = given,
                          within = TRUE,
                          vars = "M",
                          N = 1,
                          condition = "X==1 & Y==1")
make_possible_data_single(model,
                          given = given,
                          within = TRUE,
                          vars = "M",
                          N = 1,
                          condition = "X==1")

lilymedina/gbiqqtools documentation built on Nov. 4, 2019, 4:32 p.m.