Skip to main content

No project description provided

Project description

Python Package Index


cir is a python package provided the algorithm for contrastive inverse regression (CIR) for dimension reduction used in a supervised setting to find a low-dimensional representation by solving a nonconvex optimization problem on the Stiefel manifold.


Make sure you have numpy, pandas, and scipy install beforehand and the version of these packages compatible with cir. The easy way to install is using pip:

pip install contrastive-inverse-regression

Alternatively, you can also install by cloning this repository:

pip install git+


  • Python (>= 3.10.9)
  • numpy (>= 1.24.3)
  • pandas (>= 2.1.4)
  • scipy (>= 1.9.3)

To run exmaple, matplotlib (>= 3.8.2) is required


The dataset for the following example is included in the example/dataset folder.

import contrastive_inverse_regression
from contrastive_inverse_regression import CIR
import pandas as pd

d = 2
alpha = 0.0001

# download the dataset and set it to the absolute path in your computer 
fg = pd.read_csv('../foregroundX.csv')
bg = pd.read_csv('../backgroundX.csv')
Y = pd.read_csv('../foregroundY.csv')
Yt = pd.read_csv('../backgroundY.csv')

fg = fg.iloc[0:, 1:]
bg = bg.iloc[0:, 1:]
Y = Y.iloc[0:, 1:]
Yt = Yt.iloc[0:, 1:]

V = CIR(fg, Y, bg, Yt, alpha, d)

Other detailed examples for employing cir are provided.

For the case of discrete foreground Y values, the mouse protein dataset Data_Cortex_Nuclear.csv is used and the corresponding visualization in and regression testing in

For the case of continuous foreground Y values, cir is applied on the retinol dataset Retinol.txt and the corresponding regression is in Continuous values are not usually for classification, hence visualization is not provided.

Citing contrastive-inverse-regression

If you find this algorithm helpful in your research, please add the following bibtex citation in references.

  title={Contrastive inverse regression for dimension reduction},
  author={Hawke, Sam and Luo, Hengrui and Li, Didong},
  journal={arXiv preprint arXiv:2305.12287},


.. [1] : Hawke, S., Luo, H., & Li, D. (2023) "Contrastive Inverse Regression for Dimension Reduction", Retrieved from

.. [2] Harry Oviedo (2024). SGPM for minimization over the Stiefel Manifold (, MATLAB Central File Exchange. Retrieved January 12, 2024.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

contrastive_inverse_regression-1.0.13.tar.gz (7.4 kB view hashes)

Uploaded Source

Built Distribution

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page