C-Tarone: Finding Statistically Significant Interactions between Continuous Features

Source code of C-Tarone, which finds statistically significant multiplicative feature interactions from multivariate data.
Please see the following paper for more details:

• Sugiyama, M., Borgwardt, K.: Finding Statistically Significant Interactions between Continuous Features, IJCAI-19, 3490-3498 (2019).

Usage

C-Tarone is written in C++11. To compile it, the Boost library is required.

For example, in the directory src/cc:

$ make
$ ./ctarone -i synth_N=1000_n=20.dat -c synth_N=1000_05.class -o out -t stat
> Reading a database file     "synth_N=1000_n=20.dat" ... end
> Reading a class file        "synth_N=1000_05.class" ... end
  Sample size in total:       1000
  Sample size in class 0:     500
  # features:                 20
> Start enumeration of testable combinations ... end
  # testable combinations:    37910
  Corrected alpha:            1.31891e-06
  Frequency threshold:        0.0166722
  Running time:               0.456451 [sec]
> Find significant combinations with a threshold 0.0166722
  # significant combinations: 5302
  Running time:               0.490904 [sec]

Arguments

• -i <input_file>: a path to a csv file of an input dataset (without row and column names)
• -c <class_file>: a path to a file of input class labels (each line represents a binary (0/1) label of the corresponding line in <input_file>)
• -o <output_file>: Output of significant feature combinations is written to <output_file>
• -t <output_stat_file>: Output of statistics is written to <output_stat_file>
• -a <alpha>: significance level for the FWER, where C-Tarone always guarantees "FWER < <alpha>" (default: 0.05)
• -k <size_limit>: the upper bound of the size of feature combinations (default: unlimited)
• -v: Verbose mode if specified

Contact

Author: Mahito Sugiyama
Affiliation: National Institute of Informatics, Tokyo, Japan
E-mail: mahito@nii.ac.jp