Computer Science‎ > ‎

Python for Data: (12) Neural Networks in Python

Get the Data

#using pandas to read .csv file
import pandas as pd
import the data from CSV file
data = pd.read_csv('bank_note_data.csv')

    When using Neural Network and Deep Learning based systems, it is usually a good idea to Standardize your data, this step isn't actually necessary for our particular data set, but let's run through it for practice!
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()'Class',axis=1))
scaled_features = scaler.fit_transform(data.drop('Class',axis=1))

See how our data looks like after feature scaling.
X= pd.DataFrame(scaled_features,columns=data.columns[:-1])
y = data['class']
    (1372, 4)
X = X.as_matrix()
y = y.as_matrix()
    Use the .as_matrix() method on X and Y and reset them equal to this result. We need to do this in order for TensorFlow to accept the data in Numpy array form instead of a pandas series.

Train Test splitting the Data.

from sklearn.cross_validation import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3)


importing tensorflow.contrib.learn as learn

import tensorflow.contrib.learn.python.learn as learn

feature_columns = learn.infer_real_valued_columns_from_input(X)
[_RealValuedColumn(column_name='', dimension=4, default_value=None, dtype=tf.float64, 
 Creating an object called classifier which is a DNNClassifier from learn. Setting it to have 2 classes and a [10,20,10] hidden unit layer structure.

classifier = learn.DNNClassifier(hidden_units=[10, 20, 10], n_classes=2)

Fitting data to classifier and make prediction for X_test

Fitting the data to the classifier. Use steps 200 with batch_size of 20. You can play around with these values depending upon your machine limits.
    Note: Ignore any warnings you get, they won't affect your output, y_train, steps=200, batch_size=20)
note_predictions = classifier.predict(X_test)

Model Evaluation

import metrics 

from sklearn.metrics import classification_report,confusion_matrix
from numpy import array
pre = array( list(note_predictions))

    [[218   0]
     [  0 194]]
             precision    recall  f1-score   support

          0       1.00      1.00      1.00       218
          1       1.00      1.00      1.00       194

avg / total       1.00      1.00      1.00       412