Allow specification of an anchor point in time (max year)

wosis/analysis/stats.py

@@ -1,6 +1,6 @@
 __all__ = ['calc_average_citations']
 
-def calc_average_citations(in_df, sort=False):
+def calc_average_citations(in_df, max_year=None, sort=False):
     """Calculate the average citations since year of publication.
 
     Example
@@ -14,7 +14,8 @@ def calc_average_citations(in_df, sort=False):
 
     Parameters
     ==========
-    * df : Pandas DataFrame
+    * in_df : Pandas DataFrame
+	* max_year: int or None, year to calculate average citations from.
 
     Returns
     ==========
@@ -23,7 +24,15 @@ def calc_average_citations(in_df, sort=False):
     assert hasattr(in_df, 'citations'), \
         'DataFrame has to have `citation` column. Use `get_num_citations()` first'
     out_df = in_df.copy()
-    max_year = out_df.year.max()
+
+	max_year_in_data = out_df.year.max()
+
+	if max_year not None:
+		max_year = int(max_year)
+		assert max_year_in_data <= max_year, \
+			"Given max_year must be later than any year found in dataset."
+    else:
+		max_year = out_df.year.max()
     out_df.loc[:, 'Avg. Citations'] = (out_df.citations / ((max_year - out_df.year) + 1)).astype(float).round(2)
 
     if sort: