add some computation functions for groupby procedure

Author: semio

ddf_utils/ops.py

@@ -0,0 +1,53 @@
+# -*- coding: utf-8 -*-
+
+"""commonly used calculation methods"""
+
+import pandas as pd
+import numpy as np
+
+
+# Groupby Aggrgate
+
+
+# Groupby Transform
+def zcore(x):
+    return (x - x.mean()) / x.std()
+
+
+# Groupby Filter
+def gt(x, val, how='all', include_eq=False):
+    f = getattr(np, how)
+    if include_eq:
+        return f(x >= val)
+    else:
+        return f(x > val)
+
+
+def lt(x, val, how='all', include_eq=False):
+    f = getattr(np, how)
+    if include_eq:
+        return f(x <= val)
+    else:
+        return f(x <= val)
+
+
+def between(x, upper, lower, how='all', include_upper=False, include_lower=False):
+    return gt(x, lower, how, include_lower) and lt(x, upper, how, include_upper)
+
+
+# Rolling
+def aagr(df: pd.DataFrame, window: int=10):  # TODO: don't include the window
+    """average annual growth rate
+
+    Parameters
+    ----------
+    window : `int`
+        the rolling window size
+
+    Returns
+    -------
+    return : `DataFrame`
+        The rolling apply result
+    """
+    pct = df.pct_change()
+    return pct.rolling(window).apply(np.mean).dropna()

ddf_utils/transformer.py

@@ -166,23 +166,6 @@ def translate_header(df, dictionary, dictionary_type='inline'):
         raise ValueError('dictionary not supported: '+dictionary_type)
 
 
-def aagr(df: pd.DataFrame, window: int=10):  # TODO: create a op.py file for this kind of functions?
-    """average annual growth rate
-
-    Parameters
-    ----------
-    window : `int`
-        the rolling window size
-
-    Returns
-    -------
-    return : `DataFrame`
-        The rolling apply result
-    """
-    pct = df.pct_change()
-    return pct.rolling(window).apply(np.mean).dropna()
-
-
 def trend_bridge(old_data, new_data, bridge_length):
     """smoothing data between series.