DAY 19-100 DAYS MLCODE: DECISION TREES classifier
In the previous blog, we completed the decision trees. In this blog, we will create a Decision tree classifier using the Decision tree and will use
Lets create the Moon Dataset using the make_moons class of SciKit Learn
from sklearn.datasets import make_moons #Import library
X, y = make_moons(n_samples=10000, noise=0.35, random_state=42) #generate 10000 data instance
Once data is loaded , let’s split the data into train and test set
from sklearn.model_selection import train_test_split #Import train_test_split call to devidie the data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.20, random_state = 42) #Split the data
Print the size of training and testing data
print(f"Training data: {len(X_train)}, Training labels: {len(y_train)}, Testing data: {len(X_test)}, Testing labels: {len(y_test)}")
Training data: 8000, Training labels: 8000, Testing data: 2000, Testing labels: 2000
Visualize the data
Let’s plot the training data vs training labels. X-axis if the first feature and Y axis is 2nd features.
import matplotlib.pyplot as plt
plt.title("Moon Data set")
plt.scatter(X_train[:, 0], X_train[:, 1], c=y_train,
edgecolors='k')
Lets train the decision trees classifier
from sklearn.tree import DecisionTreeClassifier #Import the Decision tree classifier class
dtree_clf = DecisionTreeClassifier() #create model
dtree_clf.fit(X_train, y_train) #Train the model
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=None, splitter=’best’)
Let’s plot the decision boundary of the classifier
import numpy as np
from matplotlib.colors import ListedColormap
x_min, x_max = X[:, 0].min() - .5, X[:, 0].max() + .5
y_min, y_max = X[:, 1].min() - .5, X[:, 1].max() + .5
xx, yy = np.meshgrid(np.arange(x_min, x_max, 0.02),
np.arange(y_min, y_max, 0.02))
figure = plt.figure(figsize=(10, 10))
cm = plt.cm.RdBu
cm_bright = ListedColormap(['#FF0000', '#0000FF'])
# Plot the training points
plt.scatter(X_train[:, 0], X_train[:, 1], c=y_train, cmap=cm_bright,
edgecolors='k')
# Plot the testing points
plt.scatter(X_test[:, 0], X_test[:, 1], c=y_test, cmap=cm_bright, alpha=0.6,
edgecolors='k')
Z = dtree_clf.predict_proba(np.c_[xx.ravel(), yy.ravel()])[:, 1]
Z = Z.reshape(xx.shape)
plt.contourf(xx, yy, Z, cmap=cm, alpha=.2)
Use the Grid Search to find the best hyper-parameters using GridSearchCV class of SciKit-Learn. Let’s use the Max_Leaf_reang from 2 to 100 and min_samples_split’: [2, 3, 4] .
from sklearn.model_selection import GridSearchCV
params = {'max_leaf_nodes': list(range(2, 100)), 'min_samples_split': [2, 3, 4]}
grid_search_cv = GridSearchCV(dtree_clf , params, n_jobs=-1, verbose=1)
grid_search_cv.fit(X_train, y_train)
Fitting 3 folds for each of 294 candidates, totalling 882 fits
[Parallel(n_jobs=-1)]: Done 882 out of 882 | elapsed: 6.6s finished
GridSearchCV(cv=None, 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=None, splitter='best'), fit_params=None, iid=True, n_jobs=-1, param_grid={'max_leaf_nodes': [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99], 'min_samples_split': [2, 3, 4]}, pre_dispatch='2*n_jobs', refit=True, return_train_score='warn', scoring=None, verbose=1)
Let’s find the best estimated model parameters:
grid_search_cv.best_estimator_
DecisionTreeClassifier(class_weight=None, criterion=’gini’, max_depth=None, max_features=None, max_leaf_nodes=22, 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=None, splitter=’best’)
Since Grid search already trained the model with
Accuracy
Let’s measure the performance of our Model
from sklearn.metrics import accuracy_score #Import Librady
y_pred = grid_search_cv.predict(X_test) # Run the prediction
accuracy_score(y_test, y_pred) #Measure accuracy
Output: 0.895
Our model has predicted with accuracy of 89.5. This does not looks bad.
In conclusion, this blog has a simple implementation of decision tree classifier using SciKit-Learn. You can find the code here.