PR by harishkandan

README.md

@@ -32,4 +32,4 @@ Which country had the biggest difference between their summer and winter gold me
 Write a function to update the dataframe to include a new column called "Points" for Games which is a weighted value where each gold medal counts for 3 points, silver medals for 2 points, and bronze
 medals for 1 point. The function should return only the column (a Series object) which you created.
 * Create a function get_points.
-* Return dataframe with points column and index.
+* Return dataframe with points column and index.

build.py

@@ -1,37 +1,32 @@
 import pandas as pd
-
+import re
 
 def load_data():
-    """
-        Enter your code here
-    """
 
+    olympics=pd.read_csv('files/olympics.csv',skiprows=1)
+    olympics.rename(columns={'Unnamed: 0':'Country_name'},inplace= True)
+    for i in olympics.columns:
+        if bool(re.match(".*01 !.*",i)):
+            olympics.rename(columns={i:re.sub('01 !','Gold',i)},inplace=True)
+        elif bool(re.match(".*02 !.*",i)):
+            olympics.rename(columns={i:re.sub('02 !','Silver',i)},inplace=True)
+        elif bool(re.match(".*03 !.*",i)):
+            olympics.rename(columns={i:re.sub('03 !','Bronze',i)},inplace=True)
+    olympics['Country_name']=olympics['Country_name'].apply(lambda x:x.split("\xc2\xa0")[0])
+    olympics.index=olympics['Country_name']
+    olympics=olympics[olympics.Country_name != "Totals"]
+    return olympics
 
 def first_country(df):
-    """
-        Enter your code here
-    """
-
+    return df.iloc[0,:]
 
 def gold_medal(df):
-    """
-        Enter your code here
-    """
+    return df['Country_name'][df['Gold.2']==df['Gold.2'].max()][0]
 
 
 def biggest_difference_in_gold_medal(df):
-    """
-        Enter your code here
-    """
-
+    return df['Country_name'][abs(df['Total']-df['Total.1'])==abs(df['Total']-df['Total.1']).max()][0]
 def get_points(df):
-    """
-    Enter your code here
-    """
-
-
-# df = load_data()
-# print(first_country(df)["# Summer"])
-# print(gold_medal(df))
-# print(biggest_difference_in_gold_medal(df))
-# print(get_points(df))
+    df['points'] = (df['Gold.2']*3+df['Silver.2']*2+df['Bronze.2'])
+    return df['points']
+print biggest_difference_in_gold_medal(load_data())