Created
August 11, 2017 22:28
-
-
Save iandewancker/e698014cdfdfa2ba562a479b75fc9766 to your computer and use it in GitHub Desktop.
Comparing decision function surfaces
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
%matplotlib | |
import matplotlib | |
matplotlib.rcParams.update({'font.size': 32}) | |
import numpy as np | |
import matplotlib.pyplot as plt | |
import seaborn as sns | |
import numpy as np; np.random.seed(10) | |
mean1, cov1 = [0, 2], [(0.5, .25), (.25, 0.5)] | |
mean2, cov2 = [2.5, -0.5], [(0.01, 0.02), (0.02, 0.01)] | |
mean3, cov3 = [3.2, 1.0], [(0.01, 0.02), (0.02, 0.01)] | |
x1, y1 = np.random.multivariate_normal(mean1, cov1, size=4000).T | |
x2, y2 = np.random.multivariate_normal(mean2, cov2, size=5).T | |
x3, y3 = np.random.multivariate_normal(mean3, cov3, size=5).T | |
x = np.hstack([x1,x2,x3]) | |
y = np.hstack([y1,y2,y3]) | |
plt.figure() | |
plt.xlim(-4.0,4.0) | |
plt.ylim(-2.0,6.0) | |
plt.axes().set_aspect('equal') | |
plt.axes().set_adjustable('datalim') | |
sns.regplot(x, y,fit_reg=False, color='green') | |
plt.ylabel('Feature 1') | |
plt.xlabel('Feature 2') | |
plt.title('Samples from True Distribution') | |
x_false = np.random.uniform(-4.0,4.0,12000) | |
y_false = np.random.uniform(-2.0,6.0,12000) | |
X = np.vstack([zip(x,y),zip(x_false,y_false)]) | |
y_train = np.hstack([np.ones(y.shape[0]),np.zeros(y_false.shape[0])]) | |
from sklearn import linear_model | |
clf_lr = linear_model.LogisticRegression() | |
clf_lr.fit(X,y_train) | |
x_m = np.arange(-4.0,4.0,0.01) | |
y_m = np.arange(-2.0,6.0,0.01) | |
xm1, ym1 = np.meshgrid(x_m, y_m) | |
z_m = clf_lr.predict_proba(zip(xm1.flatten(),ym1.flatten()))[:,1] | |
z_m = z_m.reshape(xm1.shape[0], xm1.shape[1]) | |
plt.figure() | |
plt.contourf(xm1,ym1,z_m,50,cmap="BuGn") | |
plt.ylabel('Feature 1') | |
plt.xlabel('Feature 2') | |
plt.xlim((-4,4)) | |
plt.ylim((-2,6)) | |
plt.axis('equal') | |
plt.title('LR Model') | |
import sklearn.ensemble | |
clf_rf = sklearn.ensemble.RandomForestClassifier(n_estimators=10,max_depth=15) | |
clf_rf.fit(X,y_train) | |
x_m = np.arange(-4.0,4.0,0.01) | |
y_m = np.arange(-2.0,6.0,0.01) | |
xm1, ym1 = np.meshgrid(x_m, y_m) | |
z_m = clf_rf.predict_proba(zip(xm1.flatten(),ym1.flatten()))[:,1] | |
z_m = z_m.reshape(xm1.shape[0], xm1.shape[1]) | |
plt.contourf(xm1,ym1,z_m,50,cmap="BuGn") | |
plt.ylabel('Feature 1') | |
plt.xlabel('Feature 2') | |
plt.xlim((-4,4)) | |
plt.ylim((-2,6)) | |
plt.axis('equal') | |
plt.title('RF Model') | |
plt.show() |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment