changes to class

setup.py

@@ -1,9 +1,10 @@
 # -*- coding: utf-8 -*-
 from distutils.core import setup
+
 setup(
     name              = 'stepwisereg',
     packages          = ['stepwisereg'],
-    version           = '0.1.11',
+    version           = '0.1.0',
     description       = 'Stepwise Regression in Python',
     long_description  = 'Forward Stepwise Regression in Python like R using AIC',
     author = 'Avinash Barnwal',
@@ -12,8 +13,8 @@
     maintainer='Avinash Barnwal',
     maintainer_email='[email protected]',
     install_requires=[
-        'numpy',
-        'pandas',
+    'numpy',
+    'pandas',
 	'sklearn',
 	'patsy',
 	'statsmodel',

stepwisereg/example/lusc_lung_cancer.py

@@ -1,26 +1,12 @@
-import numpy as np
-import sklearn
-import warnings
-import os
-warnings.filterwarnings('ignore')
-from sklearn import datasets, linear_model
-from sklearn.metrics import mean_squared_error, r2_score
-from sklearn.datasets import make_regression
-from patsy import dmatrices
-import statsmodels.formula.api as smf
 import pandas as pd
-import functools
-import re
+from stepwisereg import *
 
-
-path = '/data_path'
-os.chdir(path)
-
-data = pd.read_csv("test_data.csv")
+data = pd.read_csv("lung_cancer.csv")
 ###########Train Dataset and Test Dataset Creation########
+
 msk = np.random.rand(len(data)) < 0.8
 train = data[msk]
-test = data[~msk]
+test  = data[~msk]
 
 #########Independent Variables are from 2:102 and 106 has Dependent Variable########
 X_train = data.iloc[:,2:102]
@@ -37,7 +23,7 @@
 
 columns_test         = list(X_test.columns.values)
 columns_changes_test = map(lambda x:x.replace("-", "_"),columns_test)
-X_test.columns = columns_changes_test
+X_test.columns       = columns_changes_test
 test = pd.concat([X_test,Y_test],axis=1)
 
 ##Creating the features concatenation
@@ -48,72 +34,8 @@
 null = 'OS_MONTHS ~' + var1
 full = 'OS_MONTHS ~' + features
 
-
-# In[10]:
-
-def reduce_concat(x, sep=""):
-    return functools.reduce(lambda x, y: str(x) + sep + str(y), x)
-
-def forward_selected(data,null_formula,full_formula,response,step,intercept):
-    """Linear model designed by forward selection.
-    Parameters:
-    -----------
-    data : pandas DataFrame with all possible predictors and response
-    response: string, name of response column in data
-    Returns:
-    --------
-    model: an "optimal" fitted statsmodels linear model
-           with an intercept
-           selected by forward selection
-           evaluated by bic
-    """
-    print(null_formula)
-    print(full_formula)
-    print(response)
-
-    null_temp        = re.split('~',null_formula)
-    null_predic_com  = null_temp[1].split('+')
-    null_predic      = null_predic_com[1:len(null_predic_com)]
-    full_temp        = re.split('~',full_formula)
-    full_predic_com  = full_temp[1].split('+')
-    full_predic      = full_predic_com[1:len(full_predic_com)]
-    indices          = [i for i,id in enumerate(full_predic) if id not in null_predic]
-    domain           = [full_predic[i] for i in indices]
-
-    start            = set(null_predic)
-    remaining        = set(domain)
-    selected         = null_predic
-    current_score, best_new_score = 10000000, 10000000
-    score_bic        = []
-    variable_added   = []
-    flag=0
-    step=2
-    while (remaining and current_score == best_new_score and step >0):
-        scores_with_candidates = []
-        for candidate in remaining:
-            formula = "{} ~ {}".format(response,' + '.join(selected + [candidate]))
-            if intercept ==1:
-                formula = formula + "-1"
-            score = smf.ols(formula, data).fit().aic
-            scores_with_candidates.append((score, candidate))
-        scores_with_candidates.sort()
-        best_new_score, best_candidate = scores_with_candidates.pop(0)
-        if current_score > best_new_score:
-            remaining.remove(best_candidate)
-            selected.append(best_candidate)
-            score_bic.append(best_new_score)
-            variable_added.append(best_candidate)
-            current_score = best_new_score
-        step=step-1
-    formula = "{} ~ {}".format(response,' + '.join(selected))
-    if intercept ==1:
-        formula = formula + "-1"
-    model = smf.ols(formula, data).fit()
-    return model
-
-model           = forward_selected(train,null,full,'OS_MONTHS',50,0)
-
-model_param     = model.params
-print(model.summary())
-
-test_predict = model.predict(test)
+model           = stepwisereg(100,1)
+model_fit       = model.fit(train,null,full,'OS_MONTHS')
+model_param     = model_fit.params
+test_predict    = model_fit.predict(test)
+print(model_fit.summary())

stepwisereg/stepwisereg.py

@@ -17,13 +17,14 @@
 
 class stepwisereg:
 
-    def __init__(self):
-        pass
+    def __init__(self,step,fit_intercept):
+        self.step = step
+        self.fit_intercept = fit_intercept
 
     def reduce_concat(self,x, sep=""):
         return functools.reduce(lambda x, y: str(x) + sep + str(y), x)
 
-    def forward_selected(self,data,null_formula,full_formula,response,step,intercept):
+    def fit(self,data,null_formula,full_formula,response):
 
         """Linear model designed by forward selection.
         Parameters:
@@ -46,20 +47,18 @@ def forward_selected(self,data,null_formula,full_formula,response,step,intercept
         full_predic      = full_predic_com[1:len(full_predic_com)]
         indices          = [i for i,id in enumerate(full_predic) if id not in null_predic]
         domain           = [full_predic[i] for i in indices]
-
         start            = set(null_predic)
         remaining        = set(domain)
         selected         = null_predic
-        current_score, best_new_score = 10000000, 10000000
-        score_bic        = []
+        current_score, best_new_score = float('inf'), float('inf')
+        score_selected   = []
         variable_added   = []
-        flag=0
-        step=2
+
         while (remaining and current_score == best_new_score and step >0):
             scores_with_candidates = []
             for candidate in remaining:
                 formula = "{} ~ {}".format(response,' + '.join(selected + [candidate]))
-                if intercept ==1:
+                if self.fit_intercept == 0:
                     formula = formula + "-1"
                 score = smf.ols(formula, data).fit().aic
                 scores_with_candidates.append((score, candidate))
@@ -68,12 +67,12 @@ def forward_selected(self,data,null_formula,full_formula,response,step,intercept
             if current_score > best_new_score:
                 remaining.remove(best_candidate)
                 selected.append(best_candidate)
-                score_bic.append(best_new_score)
+                score_selected.append(best_new_score)
                 variable_added.append(best_candidate)
                 current_score = best_new_score
             step=step-1
         formula = "{} ~ {}".format(response,' + '.join(selected))
-        if intercept ==1:
+        if self.fit_intercept == 0:
             formula = formula + "-1"
         model = smf.ols(formula, data).fit()
         return model