From 44d30a6f8db89b22773651c5d1a3e3a7638e9fbf Mon Sep 17 00:00:00 2001
From: lisyarus <lisyarus@gmail.com>
Date: Sun, 2 Jun 2024 15:46:56 +0300
Subject: [PATCH] Add fake percentile computation in util::statistics_lite
 based on normal distribution

---
 libs/util/include/psemek/util/statistics.hpp | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/libs/util/include/psemek/util/statistics.hpp b/libs/util/include/psemek/util/statistics.hpp
index 014870e4..ab2a964e 100644
--- a/libs/util/include/psemek/util/statistics.hpp
+++ b/libs/util/include/psemek/util/statistics.hpp
@@ -1,5 +1,7 @@
 #pragma once
 
+#include <boost/math/special_functions/erf.hpp>
+
 #include <iostream>
 #include <cmath>
 #include <limits>
@@ -52,6 +54,8 @@ namespace psemek::util
 		T min() const { return min_; }
 		T max() const { return max_; }
 
+		T percentile(double p) const;
+
 		template <typename H>
 		friend statistics_lite<H> merge(statistics_lite<H> const & s1, statistics_lite<H> const & s2);
 
@@ -93,6 +97,15 @@ namespace psemek::util
 		return os;
 	}
 
+	template <typename T>
+	T statistics_lite<T>::percentile(double p) const
+	{
+		// Assume normal distribution
+		// TODO: use a better distribution, maybe maximizing entropy on [0, +inf)
+		// See https://en.wikipedia.org/wiki/Differential_entropy#Alternative_proof
+		return boost::math::erf_inv(2.0 * p - 1) * var() * std::sqrt(2.0) + mean();
+	}
+
 	template <typename T>
 	statistics_lite<T> merge(statistics_lite<T> const & s1, statistics_lite<T> const & s2)
 	{