Iris TensorFlow

This notebook illustrates using tensorflow for classifying the iris dataset.

Loading The Data

First we load the dataset into a pandas.DataFrame.



In [3]:

    
import pandas as pd
import numpy as np
from sklearn import datasets
from sklearn.model_selection import KFold, cross_val_score
from ibex.tensorflow.contrib.keras.wrappers.scikit_learn import KerasClassifier as PdKerasClassifier
import tensorflow
import seaborn as sns
sns.set_style('whitegrid')

from ibex.sklearn import ensemble as pd_ensemble

%pylab inline









    



Populating the interactive namespace from numpy and matplotlib



In [2]:

    
iris = datasets.load_iris()
features = iris['feature_names']
iris = pd.DataFrame(
    np.c_[iris['data'], iris['target']], 
    columns=features+['class'])
iris.head()









    Out[2]:







  
    
      
      sepal length (cm)
      sepal width (cm)
      petal length (cm)
      petal width (cm)
      class
    
  
  
    
      0
      5.1
      3.5
      1.4
      0.2
      0.0
    
    
      1
      4.9
      3.0
      1.4
      0.2
      0.0
    
    
      2
      4.7
      3.2
      1.3
      0.2
      0.0
    
    
      3
      4.6
      3.1
      1.5
      0.2
      0.0
    
    
      4
      5.0
      3.6
      1.4
      0.2
      0.0

Building And Operating The Estimator

As usual with tensorflow/keras, We need to write a function building a model.



In [5]:

    
def buid_nn():
    np.random.seed(7)
    model = tensorflow.contrib.keras.models.Sequential()
    model.add(tensorflow.contrib.keras.layers.Dense(8, input_dim=4, activation='relu'))
    model.add(tensorflow.contrib.keras.layers.Dense(3, activation='softmax'))
    model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
    return model

buid_nn()









    Out[5]:





<tensorflow.contrib.keras.python.keras.models.Sequential at 0x11f895f28>

Now we build a PdKerasClassifier. Note the use of classes: we need to list the classes of the dependent variable.



In [6]:

    
estimator = PdKerasClassifier(
    build_fn=buid_nn, 
    classes=iris['class'].unique(), 
    epochs=200, 
    batch_size=5, 
    verbose=0)

Following sklearn conventsions, following a call to fit, the History object describing the fit is accessible via a history_ attribute.



In [9]:

    
estimator.fit(iris[features], iris['class']).history_









    Out[9]:





<tensorflow.contrib.keras.python.keras.callbacks.History at 0x120799ba8>

Evaluating Performance



In [10]:

    
kfold = KFold(n_splits=10, shuffle=True)

scores = cross_val_score(estimator, iris[features], iris['class'], cv=kfold)
scores









    Out[10]:





array([ 1.        ,  1.        ,  1.        ,  1.        ,  0.93333333,
        0.93333333,  1.        ,  0.93333333,  0.93333333,  1.        ])



In [12]:

    
sns.boxplot(x=scores, color='grey', orient='v');
ylabel('classification score (mismatch)')
figtext(
    0, 
    -0.1, 
    'Classification scores for tensorflow/keras on the Iris dataset.');



In [ ]:

	sepal length (cm)	sepal width (cm)	petal length (cm)	petal width (cm)
0	5.1	3.5	1.4	0.2
1	4.9	3.0	1.4	0.2
2	4.7	3.2	1.3	0.2
3	4.6	3.1	1.5	0.2
4	5.0	3.6	1.4	0.2

	sepal length (cm)	sepal width (cm)	petal length (cm)	petal width (cm)
0	5.1	3.5	1.4	0.2
1	4.9	3.0	1.4	0.2
2	4.7	3.2	1.3	0.2
3	4.6	3.1	1.5	0.2
4	5.0	3.6	1.4	0.2

	sepal length (cm)	sepal width (cm)	petal length (cm)	petal width (cm)
0	5.1	3.5	1.4	0.2
1	4.9	3.0	1.4	0.2
2	4.7	3.2	1.3	0.2
3	4.6	3.1	1.5	0.2
4	5.0	3.6	1.4	0.2