From bb368fe75f67adfafc5bdd0c312e86b95cbfbef9 Mon Sep 17 00:00:00 2001
From: Luke Meyers <6382230+metric-space@users.noreply.github.com>
Date: Tue, 17 Jan 2023 10:06:11 -0500
Subject: [PATCH] Doc improvement: add comment describing what convolve does
 (#264)

---
 docs/tutorials/training_agents/blackjack_tutorial.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/docs/tutorials/training_agents/blackjack_tutorial.py b/docs/tutorials/training_agents/blackjack_tutorial.py
index e91b951b3..31495887d 100644
--- a/docs/tutorials/training_agents/blackjack_tutorial.py
+++ b/docs/tutorials/training_agents/blackjack_tutorial.py
@@ -299,6 +299,7 @@ def decay_epsilon(self):
 rolling_length = 500
 fig, axs = plt.subplots(ncols=3, figsize=(12, 5))
 axs[0].set_title("Episode rewards")
+# compute and assign a rolling average of the data to provide a smoother graph
 reward_moving_average = (
     np.convolve(
         np.array(env.return_queue).flatten(), np.ones(rolling_length), mode="valid"