From f9cb9d6d42ac13064be182680666124def82b7fe Mon Sep 17 00:00:00 2001
From: Robin Windey <ro.windey@gmail.com>
Date: Sat, 27 Mar 2021 10:20:08 +0100
Subject: [PATCH] Fix doc markdown

Fixed documentation markdown remarks for
* MulticlassClassificationMetrics.LogLoss
* MulticlassClassificationMetrics.LogLossReduction

Signed-off-by: Robin Windey <ro.windey@gmail.com>
---
 .../Metrics/CalibratedBinaryClassificationMetrics.cs          | 2 +-
 .../Evaluators/Metrics/MulticlassClassificationMetrics.cs     | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/Microsoft.ML.Data/Evaluators/Metrics/CalibratedBinaryClassificationMetrics.cs b/src/Microsoft.ML.Data/Evaluators/Metrics/CalibratedBinaryClassificationMetrics.cs
index e116d94025..0c2b640187 100644
--- a/src/Microsoft.ML.Data/Evaluators/Metrics/CalibratedBinaryClassificationMetrics.cs
+++ b/src/Microsoft.ML.Data/Evaluators/Metrics/CalibratedBinaryClassificationMetrics.cs
@@ -37,7 +37,7 @@ public sealed class CalibratedBinaryClassificationMetrics : BinaryClassification
         /// <remarks>
         /// <format type="text/markdown"><![CDATA[
         /// The log-loss reduction is scaled relative to a classifier that predicts the prior for every example:
-        /// $LogLossReduction = \frac{LogLoss(prior) - LogLoss(classifier)}{LogLoss(prior)}
+        /// $LogLossReduction = \frac{LogLoss(prior) - LogLoss(classifier)}{LogLoss(prior)}$
         /// This metric can be interpreted as the advantage of the classifier over a random prediction.
         /// For example, if the RIG equals 0.2, it can be interpreted as "the probability of a correct prediction is
         /// 20% better than random guessing".
diff --git a/src/Microsoft.ML.Data/Evaluators/Metrics/MulticlassClassificationMetrics.cs b/src/Microsoft.ML.Data/Evaluators/Metrics/MulticlassClassificationMetrics.cs
index c6d2495506..ec4576b281 100644
--- a/src/Microsoft.ML.Data/Evaluators/Metrics/MulticlassClassificationMetrics.cs
+++ b/src/Microsoft.ML.Data/Evaluators/Metrics/MulticlassClassificationMetrics.cs
@@ -24,7 +24,7 @@ public sealed class MulticlassClassificationMetrics
         /// <remarks>
         /// <format type="text/markdown"><![CDATA[
         /// The log-loss metric is computed as follows:
-        /// $LogLoss = - \frac{1}{m} \sum_{i = 1}^m log(p_i),
+        /// $LogLoss = - \frac{1}{m} \sum_{i = 1}^m log(p_i)$,
         /// where $m$ is the number of instances in the test set and
         /// $p_i$ is the probability returned by the classifier
         /// of the instance belonging to the true class.
@@ -41,7 +41,7 @@ public sealed class MulticlassClassificationMetrics
         /// <remarks>
         /// <format type="text/markdown"><![CDATA[
         /// The log-loss reduction is scaled relative to a classifier that predicts the prior for every example:
-        /// $LogLossReduction = \frac{LogLoss(prior) - LogLoss(classifier)}{LogLoss(prior)}
+        /// $LogLossReduction = \frac{LogLoss(prior) - LogLoss(classifier)}{LogLoss(prior)}$
         /// This metric can be interpreted as the advantage of the classifier over a random prediction.
         /// For example, if the RIG equals 0.2, it can be interpreted as "the probability of a correct prediction is
         /// 20% better than random guessing".