From d571c4435fd02ef957000fe219ee8ee11441ccd9 Mon Sep 17 00:00:00 2001
From: Marco Cammarata <marcocamma@gmail.com>
Date: Thu, 9 Feb 2017 16:57:37 +0100
Subject: [PATCH] new module with filtering functions

---
 xray/filters.py | 85 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 85 insertions(+)
 create mode 100644 xray/filters.py

diff --git a/xray/filters.py b/xray/filters.py
new file mode 100644
index 0000000..fadabd1
--- /dev/null
+++ b/xray/filters.py
@@ -0,0 +1,85 @@
+""" 
+  module that contains filters and outliers removal procedures 
+  most of them return the data array and a dictionary with additional info
+  (parameters, statistics, etc)
+"""
+from __future__ import print_function,division
+from . import utils
+import logging
+import statsmodels.robust
+log = logging.getLogger(__name__)  # __name__ is "foo.bar" here
+
+import numpy as np
+np.seterr(all='ignore')
+
+def removeZingers(curves,errs=None,norm='auto',threshold=10,useDerivative=False):
+  """ curves will be normalized internally 
+      if errs is None, calculate mad based noise 
+      useDerivative for data with trends ..
+  """
+  # normalize
+  if norm == 'auto':
+    norm = np.nanmean(curves,axis=1)
+    norm = utils.reshapeToBroadcast(norm,curves)
+
+  if useDerivative:
+    data = np.gradient(curves/norn,axis=0)
+  else:
+    data = curves/norm
+
+  median = np.median(data,axis=0)
+
+  # calculate or normalize error
+  if errs is None:
+    errs   = statsmodels.robust.mad(data,axis=0)
+  else:
+    errs   = errs/norm
+
+  diff   = np.abs(data-median)/errs
+  idx    = diff > threshold
+  log.debug("Removed %d zingers from %d curves"%(idx.sum(),len(curves)))
+  print("Removed %d zingers from %d curves"%(idx.sum(),len(curves)))
+  if idx.sum()>0:
+    curves[idx]=np.nan
+    #curves = np.ma.MaskedArray(data=curves,mask=idx)
+  return curves
+
+def filterOutlier(curves,errs=None,norm=None,threshold=10):
+  # normalize
+  if norm == 'auto':
+    norm = np.nanmean(curves,axis=1)
+    norm = utils.reshapeToBroadcast(n,curves)
+  elif norm is None:
+    norm = 1
+
+  curves = curves/norm
+  if errs is None:
+    errs   = statsmodels.robust.mad(curves,axis=0)
+  else:
+    errs   = errs/norm
+
+  median = np.median(curves)
+  diff   = np.abs(curves-median)/errs
+  chi2   = np.sum(diff**2)/len(curves)
+  idx    = chi2 < threshold
+  return curves[idx]
+
+def chi2Filter(diffs,threshold=10):
+  """ Contrary to removeZingers, this removes entire curves """
+  idx_mask = []
+  for iscan in range(len(diffs.diffsInScanPoint)):
+    idx = diffs.chi2_0[iscan] > threshold
+    # expand along other axis (q ...)
+    #idx = utils.reshapeToBroadcast(idx,data.diffsInScanPoint[iscan])
+    idx_mask.append(idx)
+    log.debug("Chi2 mask, scanpoint: %s, curves filtereout out %d/%d (%.2f%%)"%\
+              (data.scan[iscan],idx.sum(),len(idx),idx.sum()/len(idx)*100) )
+    print("Chi2 mask, scanpoint: %s, curves filtereout out %d/%d (%.2f%%)"%\
+              (data.scan[iscan],idx.sum(),len(idx),idx.sum()/len(idx)*100) )
+
+  if "masks" not in data: data['masks'] = dict()
+  if "masks_pars" not in data: data['masks_pars'] = dict()
+  data['masks']['chi2'] = idx_mask
+  data['masks_pars']['chi2_threshold'] = threshold
+  return data
+