CovarianceWithMissing

Ignoring missing values can lead to biased estimates of the covariance.
Lounici (2012) gives an unbiased estimator when the data has missing values.

TrainFastImputation() uses training data to describe a
multivariate normal distribution that the data approximates or
can be transformed into approximating and stores this information
as an object of class 'FastImputationPatterns'. FastImputation()
function uses this 'FastImputationPatterns' object to impute (make
a good guess at) missing data in a single line or a whole data frame
of data. This approximates the process used by 'Amelia'
<https://gking.harvard.edu/amelia> but is much faster when
filling in values for a single line of data.

Stephen Haptonstahl

FastImputation

Learn from Training Data then Quickly Fill in Missing Data

CovarianceWithMissing function

<dl><dt>x</dt>
<dd>matrix or data.frame, data with each row an observation and each column a variable.</dd></dl>

Arguments

Stephen R. Haptonstahl <a href="/link/srh%40haptonstahl.org?package=FastImputation&version=2.2.1" data-mini-rdoc="FastImputation::srh@haptonstahl.org">srh@haptonstahl.org</a>

Author

Estimate covariance when data is missing — CovarianceWithMissing

<dl>

<dt>x</dt>
<dd>matrix or data.frame, data with each row an observation and each column a variable.</dd>

</dl>

Stephen R. Haptonstahl <a href='mailto:srh@haptonstahl.org'>srh@haptonstahl.org</a>

CovarianceWithMissing: Estimate covariance when data is missing

Description

Usage

Value

Arguments

Author

References