MINE Application

The mine application is a script (installed together the minepy library) which computes the MINE statistics on a comma-separated values (CSV) file. The first column of the file must contain the variable names and each variable must have the same number of samples. The file must be in the form:

var1_name,1,2.5,3,4,5,6,7,8,9,10
var2_name,8,7,6,5,6,6,6,1.23,4,4
var3_name,1,7,3,5,6,6,6,3,4,4.2
var4_name,...
...

The input file can contain missing values (new in minepy 1.1.0):

var1_name,1,2.5,3,4,5,,7,8,9,10
var2_name,8,7,,5,6,6,6,1.23,4,4
var3_name,1,7,3,,,6,6,3,4,4.2
var4_name,...
...

Only the samples without missing values are used to compute the MINE statistics. For instance, var2_name vs. var3_name:

{8,7, ,5,6,6,6,1.23,4,4  } -> {8,7,6,6,1.23,4,4  }
{1,7,3, , ,6,6,3   ,4,4.2} -> {1,7,6,6,3   ,4,4.2}

Usage:

Usage: mine infile [-a <alpha>] [-c <c>] [-o <file>] [-m <var index>] [-p <var1 index> <var2 index>]

MINE Python v. 1.1.0 [Homepage: minepy.sf.net]. The mine script
compares by default all pairs of variables against each other. It
writes an output file where each column contains MIC (strength),
MIC-r^2 (nonlinearity), MAS (non- monotonicity), MEV (functionality),
MCN (complexity, eps=0), MCN_GENERAL (complexity, eps=1-MIC) and
Pearson (r). The input must be a comma-separated values file where the
first column must contain the variable names. Each variable must have
the same numer of samples.

Options:
  -h, --help            show this help message and exit
  -a <alpha>, --alpha=<alpha>
                        the exponent in B(n) = n^alpha (default: 0.6.) alpha
                        must be in (0, 1.0]
  -c <c>, --clumps=<c>  determines how many more clumps there will be than
                        columns in every partition. Default value is 15,
                        meaning that when trying to draw Gx grid lines on the
                        x-axis, the algorithm will start with at most 15*Gx
                        clumps (default: 15). c must be > 0
  -o <file>, --output=<file>
                        output filename (default: mine_out.csv)
  -m <var index>, --master=<var index>
                        variable <var index> vs. all <var index> must be in
                        [1, number of variables in file]
  -p <var1 index> <var2 index>, --pair=<var1 index> <var2 index>
                        variable <var1 index> vs. variable <var2 index> <var1
                        index> and <var2 index> must be in [1, number of
                        variables in file]

Examples

Spellman Gene Expression dataset

Compute the MINE statistics for variable #1 (time) vs. all the other variables, with alpha=0.67 and c=15.

  1. Download the Spellman dataset

  2. From a terminal, run:

    $ mine Spellman.csv -a 0.67 -c 15 -m 1 -o Spellman_MINE.txt
    

Dataset details are in http://www.exploredata.net/Downloads

Baseball dataset

Compute the MINE statistics all pairs of variables against each other, with alpha=0.7 and c=15.

  1. Download the MLB2008 dataset

  2. From a terminal, run:

    $ mine MLB2008.csv -a 0.7 -c 15 -o MLB2008_MINE.txt
    

Dataset details are in http://www.exploredata.net/Downloads

Microbiome dataset

Compute the MINE statistics all pairs of variables against each other, with alpha=0.551 and c=10.

  1. Download the Microbiome dataset

  2. From a terminal, run:

    $ mine Microbiome.csv -a 0.551 -c 10 -o Microbiome_MINE.txt
    

Dataset details are in http://www.exploredata.net/Downloads