PR by adinathauti

build.py

@@ -1,21 +1,82 @@
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+
+df = pd.read_csv('./data/conversion_data.csv')
+#print df.head(10)
+
+
+
+
 def get_categorical_variables(df):
-    return []
+    col_names = df.columns.values
+    ls = []
+    for ele in col_names:
+        #print df[ele].value_counts().index
+        if len(df[ele].value_counts().index)<10:
+            #print("{} is a categorical column as it has only {} categories\n".format(ele,len(df[ele].value_counts().index)))
+            ls.append(ele)
+    df2 = pd.DataFrame(df,columns=ls)
+    #print df.info()
+    return df2
+
 
 
 def get_numerical_variables(df):
-    return []
+    col_names = df.columns.values
+    ls = []
+    for ele in col_names:
+        if (df[ele].dtype == 'int64' or df[ele].dtype == 'float64') & (ele not in get_categorical_variables(df).columns.values):
+            ls.append(ele)
+    df2 = pd.DataFrame(df,columns=ls)
+    return df2
+
+
 
 
 def get_numerical_variables_percentile(df):
-    pass
+    df_1 = get_numerical_variables(df)
+    dic1 =  df_1.describe().values
+    ls = zip(*dic1)
+    df_2 = pd.DataFrame(ls,columns=['count','mean','std','min','25th Percentile','50th Percentile','75th Percentile','max'])
+    ls2 = ['age','total_pages_visited']
+    df_2['variable name']=ls2
+    df_2 = df_2[['variable name','mean','25th Percentile','50th Percentile','75th Percentile']]
+    df_2['median'] = df_2['50th Percentile']
+    df_2 = df_2[['variable name','mean','median','25th Percentile','50th Percentile','75th Percentile']]
+    return df_2
 
 
 def get_categorical_variables_modes(df):
-    pass
+    df_3 = get_categorical_variables(df)
+    dic = {}
+    for ele in df_3.columns.values:
+        #print('mode of {} is {}'.format(ele,df_3[ele].mode().values[0]))
+        if ele not in dic:
+            dic[ele] = df_3[ele].mode().values[0]
+    print dic
+    df_x = pd.DataFrame(dic,index=[0],columns=dic.keys())
+    return df_x
 
 
 def get_missing_values_count(df):
-    pass
+    cols = df.columns.values
+    cnt = 0
+    df_tmp = df
+    col_ls = []
+    val_ls = []
+    for col in cols:
+        col_ls.append(col)
+        for e in df[col].values:
+            if pd.isnull(e) == True:
+                cnt+=1
+        val_ls.append(cnt)
+        cnt = 0
+    print zip(col_ls,val_ls)
+    dfz = pd.DataFrame([col_ls,val_ls])
+    dfz =  dfz.T
+    dfz.rename(columns={0:'var_name',1:'missing_value_count'},inplace=True)
+    return dfz
 
 
 def plot_histogram_with_numerical_values(df):