Computer Science‎ > ‎

Python for Data; (11) MLP Classifier in Python (Multi Layer Perceptron)


Multi-Layer Perceptron Classifier

    Let's try to solve a Kaggle Problem "Poker Rule Induction".

#import libraries
import pandas as pd

Getting the data

sampleSub = pd.read_csv("sampleSubmission.csv")
test = pd.read_csv('test.csv')
train = pd.read_csv('train.csv')
    Making our feature and prediction data seperate.
x = train.drop('hand',axis=1)
y = train['hand']
    
test.head()
x.head()
y.head()
    0    0    
    1    0
    2    2
    3    3
    4    0

Train-Test Splitting

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(x, y, test_size = 0.2)

    Definning the Model with (200,150,100) hidden layer node sizes and activation function as "relu"
from sklearn.neural_network import MLPClassifier
classifier = MLPClassifier(hidden_layer_sizes=(200,150,100),activation='relu')

Fitting the training data to the model

classifier.fit(X_train,y_train)
default parameters looks like:
MLPClassifier(activation='relu', alpha=0.0001, batch_size='auto', beta_1=0.9,
       beta_2=0.999, early_stopping=False, epsilon=1e-08,
       hidden_layer_sizes=(200, 150, 100), learning_rate='constant',
       learning_rate_init=0.001, max_iter=200, momentum=0.9,
       nesterovs_momentum=True, power_t=0.5, random_state=None,
       shuffle=True, solver='adam', tol=0.0001, validation_fraction=0.1,
       verbose=False, warm_start=False)

Prediction

pred = classifier.predict(X_test)

Model Evaluation

from sklearn.metrics import classification_report,confusion_matrix
classifier.score(X_test, y_test)

    0.92502998800479808
92% Accuracy is not that bad for starting.

print(classification_report(pred,y_test))

             precision    recall  f1-score   support

          0       0.99      0.94      0.96      2615
          1       0.93      0.93      0.93      2167
          2       0.46      0.75      0.57       141
          3       0.59      0.78      0.67        73
          4       0.08      0.40      0.13         5
          5       0.00      0.00      0.00         0
          6       0.00      0.00      0.00         0
          7       1.00      1.00      1.00         1
          8       0.00      0.00      0.00         0
          9       0.00      0.00      0.00         0

avg / total       0.94      0.93      0.93      5002


print(confusion_matrix(pred,y_test))

    [[2455  134    0    0   15   10    0    0    0    1]
     [  27 2006  111   12    8    0    0    0    1    2]
     [   0    5  106   28    0    0    2    0    0    0]
     [   0    0   13   57    0    0    3    0    0    0]
     [   1    2    0    0    2    0    0    0    0    0]
     [   0    0    0    0    0    0    0    0    0    0]
     [   0    0    0    0    0    0    0    0    0    0]
     [   0    0    0    0    0    0    0    1    0    0]
     [   0    0    0    0    0    0    0    0    0    0]
     [   0    0    0    0    0    0    0    0    0    0]]

Making Submission File

making test data suitable for prediction

Test = test.drop('id',axis=1)
Test.head()
training model with whole training data provided in the Kaggle Problem.
classifier.fit(x,y)
pred = classifier.predict(Test)
Making the submission file in the given format
pred = pd.DataFrame(pred)
sampleSub['hand'] = pred
sampleSub.to_csv('sumbit.csv',index=False)


Some of our other tutorials for Python for Data and Machine Learning