notebook.community

Edit and run



In [1]:

    
import pandas as pd
from sklearn.tree import DecisionTreeClassifier

Считываем данные



In [2]:

    
data = pd.read_csv('../data/telecom_churn.csv')

По-минимуму анализируем и предобрабатываем



In [3]:

    
data.head()









    Out[3]:







  
    
      
      State
      Account length
      Area code
      International plan
      Voice mail plan
      Number vmail messages
      Total day minutes
      Total day calls
      Total day charge
      Total eve minutes
      Total eve calls
      Total eve charge
      Total night minutes
      Total night calls
      Total night charge
      Total intl minutes
      Total intl calls
      Total intl charge
      Customer service calls
      Churn
    
  
  
    
      0
      KS
      128
      415
      No
      Yes
      25
      265.1
      110
      45.07
      197.4
      99
      16.78
      244.7
      91
      11.01
      10.0
      3
      2.70
      1
      False
    
    
      1
      OH
      107
      415
      No
      Yes
      26
      161.6
      123
      27.47
      195.5
      103
      16.62
      254.4
      103
      11.45
      13.7
      3
      3.70
      1
      False
    
    
      2
      NJ
      137
      415
      No
      No
      0
      243.4
      114
      41.38
      121.2
      110
      10.30
      162.6
      104
      7.32
      12.2
      5
      3.29
      0
      False
    
    
      3
      OH
      84
      408
      Yes
      No
      0
      299.4
      71
      50.90
      61.9
      88
      5.26
      196.9
      89
      8.86
      6.6
      7
      1.78
      2
      False
    
    
      4
      OK
      75
      415
      Yes
      No
      0
      166.7
      113
      28.34
      148.3
      122
      12.61
      186.9
      121
      8.41
      10.1
      3
      2.73
      3
      False



In [4]:

    
data.drop(['State', 'Voice mail plan'], axis=1, inplace=True)



In [5]:

    
data['International plan'] = data['International plan'].map({'Yes': 1,
                                                             'No': 0})



In [6]:

    
data.info()









    



<class 'pandas.core.frame.DataFrame'>
RangeIndex: 3333 entries, 0 to 3332
Data columns (total 18 columns):
Account length            3333 non-null int64
Area code                 3333 non-null int64
International plan        3333 non-null int64
Number vmail messages     3333 non-null int64
Total day minutes         3333 non-null float64
Total day calls           3333 non-null int64
Total day charge          3333 non-null float64
Total eve minutes         3333 non-null float64
Total eve calls           3333 non-null int64
Total eve charge          3333 non-null float64
Total night minutes       3333 non-null float64
Total night calls         3333 non-null int64
Total night charge        3333 non-null float64
Total intl minutes        3333 non-null float64
Total intl calls          3333 non-null int64
Total intl charge         3333 non-null float64
Customer service calls    3333 non-null int64
Churn                     3333 non-null bool
dtypes: bool(1), float64(8), int64(9)
memory usage: 446.0 KB



In [7]:

    
y = data['Churn'].astype('int')



In [8]:

    
X = data.drop('Churn', axis=1)



In [9]:

    
X.shape, y.shape









    Out[9]:





((3333, 17), (3333,))

Выделяем отложенную выборку (30% от исходной)



In [10]:

    
from sklearn.model_selection import train_test_split, cross_val_score
import numpy as np



In [11]:

    
X_train, X_valid, y_train, y_valid = train_test_split(X, y, 
                                                      test_size=0.3, 
                                                      random_state=17)



In [12]:

    
X_train.shape, X_valid.shape









    Out[12]:





((2333, 17), (1000, 17))

Первое дерево решений



In [13]:

    
first_tree = DecisionTreeClassifier(random_state=17)



In [14]:

    
np.mean(cross_val_score(first_tree, X_train, y_train, cv=5))









    Out[14]:





0.91427061602227722

Первая модель kNN



In [15]:

    
from sklearn.neighbors import KNeighborsClassifier



In [16]:

    
first_knn = KNeighborsClassifier()



In [17]:

    
np.mean(cross_val_score(first_knn, X_train, y_train, cv=5))









    Out[17]:





0.86712740439845226

настраиваем max_depth и max_features для дерева



In [18]:

    
from sklearn.model_selection import GridSearchCV



In [19]:

    
tree_params = {'max_depth': np.arange(1, 11), 
               'max_features':[.5, .7, 1.]}



In [20]:

    
tree_grid = GridSearchCV(first_tree, tree_params, cv=5, n_jobs=-1)



In [21]:

    
%%time
tree_grid.fit(X_train, y_train);









    



CPU times: user 2.81 s, sys: 752 ms, total: 3.56 s
Wall time: 7.45 s






    Out[21]:





GridSearchCV(cv=5, error_score='raise',
       estimator=DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=None,
            max_features=None, max_leaf_nodes=None,
            min_impurity_decrease=0.0, min_impurity_split=None,
            min_samples_leaf=1, min_samples_split=2,
            min_weight_fraction_leaf=0.0, presort=False, random_state=17,
            splitter='best'),
       fit_params=None, iid=True, n_jobs=-1,
       param_grid={'max_features': [0.5, 0.7, 1.0], 'max_depth': array([ 1,  2,  3,  4,  5,  6,  7,  8,  9, 10])},
       pre_dispatch='2*n_jobs', refit=True, return_train_score=True,
       scoring=None, verbose=0)



In [22]:

    
tree_grid.best_score_, tree_grid.best_params_









    Out[22]:





(0.94170595799399914, {'max_depth': 6, 'max_features': 1.0})

настраиваем n_neighbors для kNN



In [23]:

    
knn_params = {'n_neighbors': range(5, 30, 5) }#+ list(range(50, 100, 10))}



In [24]:

    
knn_grid = GridSearchCV(first_knn, knn_params, cv=5)



In [25]:

    
%%time
knn_grid.fit(X_train, y_train);









    



CPU times: user 5.07 s, sys: 0 ns, total: 5.07 s
Wall time: 5.05 s






    Out[25]:





GridSearchCV(cv=5, error_score='raise',
       estimator=KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
           metric_params=None, n_jobs=1, n_neighbors=5, p=2,
           weights='uniform'),
       fit_params=None, iid=True, n_jobs=1,
       param_grid={'n_neighbors': range(5, 30, 5)},
       pre_dispatch='2*n_jobs', refit=True, return_train_score=True,
       scoring=None, verbose=0)



In [26]:

    
knn_grid.best_score_, knn_grid.best_params_









    Out[26]:





(0.8701243034719246, {'n_neighbors': 10})

прогноз для отложенной выборки



In [27]:

    
tree_valid_pred = tree_grid.predict(X_valid)



In [28]:

    
from sklearn.metrics import accuracy_score



In [29]:

    
accuracy_score(y_valid, tree_valid_pred)









    Out[29]:





0.94599999999999995

отрисовка дерева



In [30]:

    
from sklearn.tree import export_graphviz



In [31]:

    
second_tree = DecisionTreeClassifier(max_depth=3).fit(X_train, y_train)
second_tree.score(X_valid, y_valid)









    Out[31]:





0.90500000000000003



In [32]:

    
export_graphviz(second_tree, out_file='../img/telecom_tree.dot',
               feature_names=X.columns, filled=True)



In [33]:

    
!dot -Tpng ../img/telecom_tree.dot -o ../img/telecom_tree.png
!rm ../img/telecom_tree.dot

	State	Account length	Area code	International plan	Voice mail plan	Number vmail messages	Total day minutes	Total day calls	Total day charge	Total eve minutes	Total eve calls	Total eve charge	Total night minutes	Total night calls	Total night charge	Total intl minutes	Total intl calls	Total intl charge	Customer service calls	Churn
0	KS	128	415	No	Yes	25	265.1	110	45.07	197.4	99	16.78	244.7	91	11.01	10.0	3	2.70	1	False
1	OH	107	415	No	Yes	26	161.6	123	27.47	195.5	103	16.62	254.4	103	11.45	13.7	3	3.70	1	False
2	NJ	137	415	No	No	0	243.4	114	41.38	121.2	110	10.30	162.6	104	7.32	12.2	5	3.29	0	False
3	OH	84	408	Yes	No	0	299.4	71	50.90	61.9	88	5.26	196.9	89	8.86	6.6	7	1.78	2	False
4	OK	75	415	Yes	No	0	166.7	113	28.34	148.3	122	12.61	186.9	121	8.41	10.1	3	2.73	3	False