From 0936ce21c99dc833f0ad2908cf8184cf435258ab Mon Sep 17 00:00:00 2001
From: Olivier Grisel <olivier.grisel@ensta.org>
Date: Thu, 2 Jan 2020 13:47:02 +0100
Subject: [PATCH] Make test_sag_regressor_computed_correctly deterministic

Fix #15818.
---
 sklearn/linear_model/tests/test_sag.py | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/sklearn/linear_model/tests/test_sag.py b/sklearn/linear_model/tests/test_sag.py
index 6a591288b55d8..6bb156c64715b 100644
--- a/sklearn/linear_model/tests/test_sag.py
+++ b/sklearn/linear_model/tests/test_sag.py
@@ -120,7 +120,7 @@ def sag(X, y, step_size, alpha, n_iter=1, dloss=None, sparse=False,
 
 def sag_sparse(X, y, step_size, alpha, n_iter=1,
                dloss=None, sample_weight=None, sparse=False,
-               fit_intercept=True, saga=False):
+               fit_intercept=True, saga=False, random_state=0):
     if step_size * alpha == 1.:
         raise ZeroDivisionError("Sparse sag does not handle the case "
                                 "step_size * alpha == 1")
@@ -130,7 +130,7 @@ def sag_sparse(X, y, step_size, alpha, n_iter=1,
     sum_gradient = np.zeros(n_features)
     last_updated = np.zeros(n_features, dtype=np.int)
     gradient_memory = np.zeros(n_samples)
-    rng = np.random.RandomState(77)
+    rng = check_random_state(random_state)
     intercept = 0.0
     intercept_sum_gradient = 0.0
     wscale = 1.0
@@ -368,7 +368,7 @@ def test_sag_regressor_computed_correctly():
     alpha = .1
     n_features = 10
     n_samples = 40
-    max_iter = 50
+    max_iter = 100
     tol = .000001
     fit_intercept = True
     rng = np.random.RandomState(0)
@@ -378,7 +378,8 @@ def test_sag_regressor_computed_correctly():
     step_size = get_step_size(X, alpha, fit_intercept, classification=False)
 
     clf1 = Ridge(fit_intercept=fit_intercept, tol=tol, solver='sag',
-                 alpha=alpha * n_samples, max_iter=max_iter)
+                 alpha=alpha * n_samples, max_iter=max_iter,
+                 random_state=rng)
     clf2 = clone(clf1)
 
     clf1.fit(X, y)
@@ -387,12 +388,14 @@ def test_sag_regressor_computed_correctly():
     spweights1, spintercept1 = sag_sparse(X, y, step_size, alpha,
                                           n_iter=max_iter,
                                           dloss=squared_dloss,
-                                          fit_intercept=fit_intercept)
+                                          fit_intercept=fit_intercept,
+                                          random_state=rng)
 
     spweights2, spintercept2 = sag_sparse(X, y, step_size, alpha,
                                           n_iter=max_iter,
                                           dloss=squared_dloss, sparse=True,
-                                          fit_intercept=fit_intercept)
+                                          fit_intercept=fit_intercept,
+                                          random_state=rng)
 
     assert_array_almost_equal(clf1.coef_.ravel(),
                               spweights1.ravel(),