reducing noise by half

1 files changed, 47 insertions, 0 deletions

diff --git a/data/combined/mean.py b/data/combined/mean.py
new file mode 100755
index 0000000..2499beb
--- /dev/null
+++ b/data/combined/mean.py
@@ -0,0 +1,47 @@
+#!/usr/bin/python
+
+import sys
+import numpy as np
+
+sk_file = open(sys.argv[1])
+try:
+    div = float(sys.argv[2])
+except:
+    div = 2.0
+labels = sk_file.readline().strip().split(',')
+beg = labels.index('h-sc')
+data = {}
+#print ','.join(['#name'] + labels[beg:])
+for i in range(1,40):
+    data[str(i)] = {}
+    for l in labels[beg:]:
+        data[str(i)][l] = []
+
+for line in sk_file:
+    line = line.strip().split(',')
+    #limbs = line[beg:]
+    if "-1" not in line and 2 < float(line[labels.index('z-value')]) < 3:
+        id = line[labels.index('name')]
+        for x in labels[beg:]: data[id][x] += [float(line[labels.index(x)])]
+
+for i in data:
+    out = [i]
+    for l in labels[beg:]:
+        if len(data[i][l]) > 1:
+            data[i][l] = np.average(data[i][l])
+            out += [str(np.average(data[i][l]))]
+    #if len(out) > 1:
+    #    print ','.join(out)
+
+sk_file.seek(0)
+print sk_file.readline().strip()
+for line in sk_file:
+    line = line.strip().split(',')
+    id = line[labels.index('name')]
+    for l in labels[beg:]:
+        obs = float(line[labels.index(l)])
+        if obs > 0:
+            diff = obs-data[id][l]
+            line[labels.index(l)] = str(obs - diff/div)
+    print ','.join(line)
+