Displaying a model matrix in R

Jaye Ross — Tue, 15 Oct 2024 17:29:37 +0000

If you want to see how R will represent a factor in a model, you can use the model.matrix function.

For example, with unordered factors the model produces dummy variables.

> p = letters[1:4] |> factor()
> model.matrix(~p)
  (Intercept) pb pc pd
1           1  0  0  0
2           1  1  0  0
3           1  0  1  0
4           1  0  0  1

For ordered factors, it produces polynomial contrasts.

> p = letters[1:4] |> factor(ordered = TRUE)
> model.matrix(~p)
  (Intercept)        p.L  p.Q        p.C
1           1 -0.6708204  0.5 -0.2236068
2           1 -0.2236068 -0.5  0.6708204
3           1  0.2236068 -0.5 -0.6708204
4           1  0.6708204  0.5  0.2236068

Original source.

How to convert 2x2 contingency tables into a format the infer package can use

Jaye Ross — Fri, 27 Sep 2024 16:41:09 +0000

Suppose you have this data (from http://www.feat.engineering/stroke-tour):

data = c(43, 39, 19, 25)

Suppose you want to perform a Chi-squared test of association. In base R, you can do this:

data |>
matrix(nrow=2) |>
chisq.test()

However, suppose you want to use the infer package. Unlike chisq.test, you cannot provide the contingency table directly. In order to use the package, you can convert the table into the appropriate format like this. The idea is to first create the combinations of the factors and then use uncount to produce the correct format:

expand.grid(
  stroke = factor(c(1,0)),
  blockage = factor(c(1,0))
) |>
bind_cols(value = data) |>
uncount(value) |>
chisq_test(stroke ~ blockage)

DEV Community: Jaye Ross

Displaying a model matrix in R

How to convert 2x2 contingency tables into a format the infer package can use