From 14247b7d1816fa73f06dd1cfc0b82e4c48b9b1ea Mon Sep 17 00:00:00 2001
From: Tom Selier <tbj.selier@gmail.com>
Date: Fri, 29 Sep 2023 17:36:56 +0200
Subject: [PATCH] added weighted std calcs

---
 .../algorithms/data/{data.csv => sift.csv}    |  0
 .../algorithms/distribution_test.py           | 84 +++++++++++++++++++
 2 files changed, 84 insertions(+)
 rename src/experiments/algorithms/data/{data.csv => sift.csv} (100%)
 create mode 100644 src/experiments/algorithms/distribution_test.py

diff --git a/src/experiments/algorithms/data/data.csv b/src/experiments/algorithms/data/sift.csv
similarity index 100%
rename from src/experiments/algorithms/data/data.csv
rename to src/experiments/algorithms/data/sift.csv
diff --git a/src/experiments/algorithms/distribution_test.py b/src/experiments/algorithms/distribution_test.py
new file mode 100644
index 0000000..8a08044
--- /dev/null
+++ b/src/experiments/algorithms/distribution_test.py
@@ -0,0 +1,84 @@
+import numpy as np
+import matplotlib.pyplot as plt
+import cv2
+import os
+import math
+
+def imgStats(img):
+    mean = np.zeros(3)
+    std = np.zeros(3)
+    for i in range(img.shape[2]):
+        mean[i] = np.mean(img[:, :, i])
+        std[i] = np.std(img[:, :, i])
+    return mean, std
+
+def isFloat(num):
+    try:
+        float(num)
+        return True
+    except ValueError:
+        return False
+    
+def calcWeightedStd(values, weights):
+    N = len(values)
+    x_bar = np.average(values, weights=weights, axis=0)
+    numerator = 0
+    for i in range(N):
+        numerator += weights[i] * ((values[i] -  x_bar) ** 2)
+    denominator = np.sum(weights)
+    return math.sqrt(numerator / denominator)
+
+def calcNormalFunc(mean, sd, len):
+    f = np.zeros(len, dtype=np.longdouble)
+    
+    # calculate PDF
+    for x in range(len):
+        exp = (-(x - mean) ** 2)/(2 * sd ** 2)
+        f[x] = 1 / math.sqrt(2 * np.pi * sd** 20 ) * (math.exp(exp))
+
+    # normalize PDF
+    max = np.amax(f)
+    min = np.amin(f)
+    for x in range(len):
+        f[x] = (f[x] - min) / (max - min)
+
+    return f
+
+DATASET_PATH = "dataset\\"
+BARK_TYPES = 8
+
+averages = [[] for x in range(BARK_TYPES)]
+weights = [[] for x in range(BARK_TYPES)] 
+variances = [[] for x in range(BARK_TYPES)] 
+           
+i = -1
+last_name = ""
+for file in os.listdir(DATASET_PATH):
+    name = file.split('_')[0]
+    if(name != last_name):
+        last_name = name
+        i += 1
+
+    image = cv2.imread(os.path.join(DATASET_PATH + file), 1)
+    assert image is not None, "Something went wrong"
+
+    # Weighted average
+    averages[i].append(np.mean(image[:, :, 0]))
+    weights[i].append(len(image[:, :, 0]))
+    
+    # print()
+
+## Voor gewogen std: 
+## https://stackoverflow.com/questions/2413522/weighted-standard-deviation-in-numpy
+weighted_avg = np.zeros(BARK_TYPES)
+weighted_std = np.zeros(BARK_TYPES)
+for i in range(BARK_TYPES):
+    weighted_avg[i] = np.average(averages[i], weights=weights[i], axis=0)
+    weighted_std[i] = calcWeightedStd(averages[i], weights[i])
+
+print("Weighted averages: ")
+print(weighted_avg)
+
+print("Weighted standard deviation: ")
+print(weighted_std)
+