Collapse a large data set for heatmap.fit

Share:

Description

Reduces the size of large binary data sets by binning them according to their predicted probability [0, 1].

Usage

1
heatmap.compress(y, pred, init.grid)

Arguments

y

A vector of observations of the dependent variable (in {0,1}).

pred

A vector of predicted Pr(y = 1) corresponding to each element of y.

init.grid

The number of bins on the interval [0, 1] to use for compression of pred.

Value

A list with the elements:

y.out

The value of y, 0 or 1.

pred.out

The (binned) predicted Pr(y = 1) matching each observation.

weight.out

A weight parameter indicating the proportion of observations in the bin; sums to one.

pred.total.out

A vector of unique Pr(y = 1) bin values.

n.out

The number of observations (non-empty bins) after the data are collapsed.

retained.obs

A vector of indices for non-empty candidate bins (for internal use by heatmap.fit).

Author(s)

Justin Esarey <justin@justinesarey.com>

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.