Reproducible code for the zero-inflated latent Ising analysis

Author: Jie Zhou

Documentation can be found at documentation folder. This is in R documentation format and can be easily accessed through RStudio.

Detailed documentation as follows:

Binary: Generate the design matrix for logistic regression

Description

This function is used to generate the design matrix from a the given categorical observation.

Usage

Binary(MultiLevel, N)

Arguments

Argument	Description
MultiLevel	refers to the categorical observations
N	vector indicating the degree of freedom for each categorical variable

MultiLevel: Transform all the variables into categorical variables with a upper bound for degree of freedom

Description

Only a common upper bound MaxGroup for the degree of freedom is required for the data transformation.

Usage

MultiLevel(data, MaxGroup, size)

Arguments

Argument	Description
data	n by p data matrix
MaxGroup	the common upper bound for the number of levels

netmat: Generate the adjacent matrix using the zero-inflated latent Ising model

Description

This is the main function that implement our algorithm.

Usage

netmat(A, width)

Arguments

Argument	Description
A	a binary matrix representing the transfomred data
width	a vector indicating which colomns are belonging to the same categorical variable

CuttingPoint: Transform the continuous data into categorical data

Description

This function uses dynamic programming algorithm to transform the zero-inflated continuous variable into categorical variable that minimize the squared Euclidean distance.

Usage

CuttingPoint(x, N)

Arguments

Argument	Description
x	vector x representing the observations on continuous variable to be transformed.
N	N representing the specified number of the levels