notebook.community

Edit and run

Classification Trees

Load file

Commonly two libraries are used to load a csv files.

  • numpy function np.loadtext and np.genfromtext
  • pandas function pd.read_csv

Here we prefer using pandas





In [1]:



    


import pandas as pd

path='/Users/vahidpartovinia/Desktop/'
# path is your local file address
# in MS Windows you must write 'C:\\Users\\..'

filename = path+'spamdata.csv'
spam = pd.read_csv(filename)



    











In [2]:



    


# see the data
spam.head()



    


















    
Out[2]:











  
    

      
      make
      address
      all
      3d
      our
      over
      remove
      internet
      order
      mail
      ...
      semicol
      parentheses
      backet
      exclamation
      dollar
      sharp
      length_average
      length_longest
      length_total
      spam
    

  
  
    

      0
      0.00
      0.64
      0.64
      0.0
      0.32
      0.00
      0.00
      0.00
      0.00
      0.00
      ...
      0.00
      0.000
      0.0
      0.778
      0.000
      0.000
      3.756
      61
      278
      1
    

    

      1
      0.21
      0.28
      0.50
      0.0
      0.14
      0.28
      0.21
      0.07
      0.00
      0.94
      ...
      0.00
      0.132
      0.0
      0.372
      0.180
      0.048
      5.114
      101
      1028
      1
    

    

      2
      0.06
      0.00
      0.71
      0.0
      1.23
      0.19
      0.19
      0.12
      0.64
      0.25
      ...
      0.01
      0.143
      0.0
      0.276
      0.184
      0.010
      9.821
      485
      2259
      1
    

    

      3
      0.00
      0.00
      0.00
      0.0
      0.63
      0.00
      0.31
      0.63
      0.31
      0.63
      ...
      0.00
      0.137
      0.0
      0.137
      0.000
      0.000
      3.537
      40
      191
      1
    

    

      4
      0.00
      0.00
      0.00
      0.0
      0.63
      0.00
      0.31
      0.63
      0.31
      0.63
      ...
      0.00
      0.135
      0.0
      0.135
      0.000
      0.000
      3.537
      40
      191
      1
    

  



5 rows × 58 columns



















In [3]:



    


#features against spam, spam=1, ham=0
import matplotlib.pyplot as plt
%matplotlib inline 

#plot column number 12, versus ('ham or spam')

plt.scatter(spam.values[:,11], spam.values[:,-1]);



    


















    
























In [4]:



    


from sklearn.tree import DecisionTreeClassifier
from sklearn.tree import export_graphviz

# you may play with max_depth  

dt = DecisionTreeClassifier(max_depth=2)
# you may play with depth and prune the tree in different levels

X = spam.values[:, :57]
y = spam.values[:, -1]

dt.fit(X,y)
spamnames = spam.columns.tolist()[:57]



    











In [5]:



    


dot_data = export_graphviz(dt, out_file=None, 
                         feature_names=spamnames,  
                         filled=True, rounded=True,  
                         special_characters=True)



    











In [6]:



    


dot_data = export_graphviz(dt, out_file=None)



    











In [7]:



    


# you need to install graphviz-python from Anaconda, 
# graphviz-python it is not installed by default.
import graphviz



    











In [8]:



    


graph = graphviz.Source(dot_data)



    











In [9]:



    


dot_data = export_graphviz(dt, out_file=None, 
                         feature_names=spamnames,  
                         class_names=['ham', 'spam'],  
                         filled=True, rounded=True,  
                         special_characters=True)  
graph = graphviz.Source(dot_data)  
graph



    


















    
Out[9]:













Tree



0

dollar ≤ 0.056
gini = 0.478
samples = 4601
value = [2788, 1813]
class = ham



1

remove ≤ 0.055
gini = 0.36
samples = 3471
value = [2655, 816]
class = ham



0->1


True



4

hp ≤ 0.4
gini = 0.208
samples = 1130
value = [133, 997]
class = spam



0->4


False



2

gini = 0.275
samples = 3141
value = [2625, 516]
class = ham



1->2





3

gini = 0.165
samples = 330
value = [30, 300]
class = spam



1->3





5

gini = 0.123
samples = 1060
value = [70, 990]
class = spam



4->5





6

gini = 0.18
samples = 70
value = [63, 7]
class = ham



4->6






















In [10]:



    


from sklearn.model_selection import train_test_split



    











In [11]:



    


X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state=1)



    











In [12]:



    


dt10 = DecisionTreeClassifier(max_depth=10)
dt11 = DecisionTreeClassifier(max_depth=11)
dt12 = DecisionTreeClassifier(max_depth=12)

# you may play with depth and prune the tree in different levels
dt10.fit(X_train,y_train)
dt11.fit(X_train,y_train)
dt12.fit(X_train,y_train)

y10_pred = dt10.predict(X_test)
y11_pred = dt11.predict(X_test)
y12_pred = dt12.predict(X_test)

from sklearn.metrics import accuracy_score



    











In [13]:



    


accuracy_score(y_test, y10_pred)



    


















    
Out[13]:








0.91639522258414763

















In [14]:



    


accuracy_score(y_test, y11_pred)



    


















    
Out[14]:








0.91965255157437564

















In [15]:



    


accuracy_score(y_test, y12_pred)



    


















    
Out[15]:








0.91313789359391961

Content source: vahidpartovinia/website

Similar notebooks:

notebook.community | gallery | about