# Multivariate Outlier Detection and Imputation for Incomplete Survey Data

Algorithms for multivariate outlier detection when missing values occur. Algorithms are based on Mahalanobis distance or data depth. Imputation is based on the multivariate normal model or uses nearest neighbour donors. The algorithms take sample designs, in particular weighting, into account. The methods are described in Bill and Hulliger (2016) .

# modi

The package modi provides several functions for multivariate outlier detection and imputation. They may be used when analysing multivariate quantitative survey data. The distribution of such data is often not multivariate normal. Futhermore, the data is often skewed and exhibits features of a semi-continuous distribution. Finally, missing values and non-response is common. The functions provided in modi address those problems.

## Overview

The following outlier detection and imputation functions are provided in modi:

• `BEM()` is an implementation of the BACON-EEM algorithm to detect outliers under the assumption of a multivariate normal distribution.
• `TRC()` is an implementation of the transformed rank correlation (TRC) algorithm to detect outliers.
• `EAdet()` is an outlier detection method based on the epidemic algorithm (EA).
• `EAimp()` is an imputation method based on the epidemic algorithm (EA).
• `GIMCD()` is an outlier detection method based on non-robust Gaussian imputation (GI) and the highly robust minimum covariance determinant (MCD) algorithm.
• `POEM()` is a nearest neighbor imputation method for outliers and missing values.
• `winsimp()` is an imputation method for outliers and missing values based on winsorization and Gaussian imputation.
• `ER()` is a robust multivariate outlier detection algorithm that can cope with missing values.

## Installation

You can install modi from github by runing the following line of code in R:

## Example

The following simple example shows how the BACON-EEM algorithm can be applied to detect outliers in the Bushfire dataset:

# Reference manual

install.packages("modi")

0.1.0 by Martin Sterchi, a year ago

https://github.com/martinSter/modi

Report a bug at https://github.com/martinSter/modi/issues

Browse source code at https://github.com/cran/modi

Authors: Beat Hulliger [aut] , Martin Sterchi [cre]

Documentation:   PDF Manual