Last active
May 26, 2021 01:48
-
-
Save dewith/6869438a600591a9be5ddc3977ab369c to your computer and use it in GitHub Desktop.
Python function to create multiple plots to evaluate the model performance visually.
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
def evaluate_model(y_train, y_train_pred, y_test, y_test_pred): | |
'''Creates multiple plots to evaluate the model performance visually. | |
Parameters | |
---------- | |
y_train: array | |
The target for the train set. | |
y_train_pred: array | |
The predicted target for the train set. | |
y_test: numpy array | |
The target for the test set. | |
y_test_pred: array | |
The predicted target for the test set. | |
Returns | |
------- | |
Matplotlib GridSpec | |
Notes | |
------ | |
Requirements: Matplotlib, Seaborn. | |
''' | |
from matplotlib.ticker import FuncFormatter | |
from matplotlib.gridspec import GridSpec | |
# This is optional, it's made for large values (i.e., money) | |
def kilos(x, pos): | |
'The two args are the value and tick position' | |
return '%3.0fK' % (x * 1e-3) | |
formatter = FuncFormatter(kilos) | |
# End of optional | |
fig = plt.figure(constrained_layout=False, figsize=(12, 9)) | |
widths = [5, 5] | |
heights = [3, 5] | |
gs = fig.add_gridspec(ncols=2, nrows=2, wspace=0.4, hspace=0.35, | |
width_ratios=widths, | |
height_ratios=heights) | |
# Histogram of Errors | |
ax1 = fig.add_subplot(gs[0, :]) | |
ax1.set_title('Histogram of Errors') | |
train_errors = y_train - y_train_pred | |
test_errors = y_test - y_test_pred | |
rango = (min(np.min(train_errors), | |
np.min(test_errors)), | |
max(np.max(train_errors), | |
np.max(test_errors))) | |
sns.distplot(train_errors, kde=False, bins=45, label='Train', | |
hist_kws={'range': rango, 'alpha': 1}, | |
ax=ax1) | |
sns.distplot(test_errors, kde=False, bins=45, label='Test', | |
hist_kws={'range': rango, 'alpha': 1}, | |
ax=ax1) | |
ax1.set_ylabel('Quantity') | |
ax1.set_xlabel('Error') | |
ax1.legend(fancybox=True, loc='right') | |
ax1.xaxis.set_major_formatter(formatter) | |
# Residuals plot | |
ax2 = fig.add_subplot(gs[1, 0]) | |
ax2.set_title('Residuals Plot') | |
ax2.scatter(y_train_pred, y_train_pred - y_train, | |
s=25, edgecolor='white', linewidths=0.5, alpha=1, | |
label='Train') | |
ax2.scatter(y_test_pred, y_test_pred - y_test, | |
s=25, edgecolor='white', linewidths=0.5, alpha=1, | |
label='Test') | |
ax2.hlines(y=0, | |
xmin=np.min(ax2.get_xlim()), | |
xmax=np.max(ax2.get_xlim()), | |
color='black', alpha=0.9, | |
lw=2) | |
ax2.set_xlabel('Predicted values') | |
ax2.set_ylabel('Residuals') | |
ax2.legend(fancybox=True, loc='lower right') | |
ax2.xaxis.set_major_formatter(formatter) | |
ax2.yaxis.set_major_formatter(formatter) | |
# Prediction errors plot | |
ax3 = fig.add_subplot(gs[1, 1]) | |
ax3.set_title('Prediction Errors Plot') | |
ax3.scatter(y_test, y_test_pred, color=red, | |
s=25, edgecolor='white', linewidths=0.5, alpha=1) | |
lims = [np.min([ax3.get_xlim(), ax3.get_ylim()]), | |
np.max([ax3.get_xlim(), ax3.get_ylim()])] | |
ax3.plot(lims, lims, '-k', alpha=1, lw=2, label='Identity') | |
sns.regplot(y_test, y_test_pred, scatter=False, ci=None, | |
color='black', label='Best fit', | |
line_kws={'ls': '--', 'alpha':0.6, 'lw':2}) | |
ax3.set_xlabel('Real (test)') | |
ax3.set_ylabel('Predicted (test)') | |
ax3.legend(fancybox=True, loc='lower right') | |
ax3.xaxis.set_major_formatter(formatter) | |
ax3.yaxis.set_major_formatter(formatter) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment