From 3ecb48057afa2dd805f7214166340af987af9b0e Mon Sep 17 00:00:00 2001
From: 0xfdf <0xfdf@implies.com>
Date: Tue, 13 Aug 2024 00:01:50 -0400
Subject: [PATCH] fix merge

---
 toraniko/math.py | 53 ------------------------------------------------
 1 file changed, 53 deletions(-)

diff --git a/toraniko/math.py b/toraniko/math.py
index fe3868a..dd1ad2d 100644
--- a/toraniko/math.py
+++ b/toraniko/math.py
@@ -238,56 +238,3 @@ def ledoit_wolf_shrinkage(X: np.ndarray) -> tuple[float | int, np.ndarray]:
     shrunk_cov = (1 - shrinkage) * sample_cov + shrinkage * mu * np.eye(m)
 
     return shrinkage, shrunk_cov
-
-
-# TODO: test
-def stfu_shrinkage(X: np.ndarray) -> tuple[tuple[float | int, ...], np.ndarray]:
-    """Estimate the covariance matrix of `X` via Specializing the Target to Features and Unlabeled shrinkage (SFTU).
-
-    Parameters
-    ----------
-    X : array-like input data matrix for which to estimate covariance, having shape (n_samples, m_features)
-
-    Returns
-    -------
-    shrinkage: float estimated shrinkage parameter.
-    shrunk_cov: numpy ndarray estimated shrunk covariance matrix having shape (n_features, n_features)
-    """
-    n, m = X.shape
-
-    # Center the data
-    X = X - X.mean(axis=0)
-
-    # Sample covariance
-    sample_cov = np.dot(X.T, X) / n
-
-    # Diagonal of sample covariance
-    sample_var = np.diag(sample_cov)
-
-    # Mean of sample variances
-    mean_var = np.mean(sample_var)
-
-    # Estimate of tr(sigma^2)
-    tr_sigma2 = np.sum(sample_cov**2)
-
-    # Estimate of tr(sigma^4)
-    tr_sigma4 = np.sum((X.T @ X / n) ** 2)
-
-    # Estimate lambda_star (shrinkage towards diagonal matrix)
-    lambda_star = (tr_sigma2 - np.sum(sample_var**2)) / (
-        (n - 1) * (tr_sigma2 - 2 * np.sum(sample_var**2) + m * mean_var**2)
-    )
-    lambda_star = max(0, min(1, lambda_star))
-
-    # Estimate mu_star (shrinkage towards scalar matrix)
-    mu_star = (tr_sigma2 - m * mean_var**2) / ((n - 1) * (tr_sigma4 - tr_sigma2**2 / m))
-    mu_star = max(0, min(1, mu_star))
-
-    # Compute shrunk covariance matrix
-    shrunk_cov = (
-        (1 - lambda_star) * sample_cov
-        + lambda_star * np.diag(sample_var)
-        + lambda_star * mu_star * (mean_var - np.diag(sample_var))
-    )
-
-    return (lambda_star, mu_star), shrunk_cov