[FIXED] Is there a way to plot multiple graphs with multiple lines on each one using matplotlib?

Issue

I'm trying to plot learning curves for different metrics such as training and validation loss and accuracy.

for lr in learning_rates: 
    opt = SGD(learning_rate = lr)
    model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])
    network = model.fit(x_train_subset, y_train_subset, epochs=iterations, batch_size=32, validation_data=(X_test, y_test), verbose=1)

    train_loss_values = network.history['loss']
    train_accuracy_values = network.history['accuracy']
    val_loss_values = network.history['val_loss']
    val_accuracy_values = network.history['val_accuracy']

    plt.plot(epochs, train_accuracy_values, label=f'LR = {lr}')
    
plt.title('Training Accuracy over Epochs for Different Learning Rates')
plt.xlabel('Epochs')
plt.ylabel('Training Accuracy')
plt.legend()
plt.grid(True)
plt.show()

This is the code I have and it works exactly how I want it to for the one graph with training accuracy as the metric.

Output of the above code:

However, I want to replicate this four times for all the different metrics stored in the following code:

train_loss_values = network.history['loss']
train_accuracy_values = network.history['accuracy']
val_loss_values = network.history['val_loss']
val_accuracy_values = network.history['val_accuracy']

If I simply do the following, it does not create 4 separate graphs, it just adds all data to one.

for lr in learning_rates: 
    opt = SGD(learning_rate = lr)
    model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])
    network = model.fit(x_train_subset, y_train_subset, epochs=iterations, batch_size=32, validation_data=(X_test, y_test), verbose=1)

    train_loss_values = network.history['loss']
    train_accuracy_values = network.history['accuracy']
    val_loss_values = network.history['val_loss']
    val_accuracy_values = network.history['val_accuracy']

    plt.plot(epochs, train_accuracy_values, label=f'LR = {lr}')
    plt.plot(epochs, train_loss_values, label=f'LR = {lr}')
    plt.plot(epochs, val_accuracy_values, label=f'LR = {lr}')
    plt.plot(epochs, val_loss_values, label=f'LR = {lr}')
    
plt.title('Training Accuracy over Epochs for Different Learning Rates')
plt.xlabel('Epochs')
plt.ylabel('Training Accuracy')
plt.legend()
plt.grid(True)
plt.show()

I asked ChatGPT and it always suggested running the entire neural network again 4 times, but I have already run it once and collected all the data I need in lists. I just need to create 4 separate plots.

Solution

You can use plt.subplot in order to display multiple plots.
plt.subplot(2,2,i) creates 4 plots in a 2 x 2 grid. i is the index of the plot you are currently setting.

#Set the size of your plot
plt.subplots(2, 2, figsize=(15, 10))

for lr in learning_rates: 
    opt = SGD(learning_rate = lr)
    model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])
    network = model.fit(x_train_subset, y_train_subset, epochs=iterations, batch_size=32, validation_data=(X_test, y_test), verbose=1)

    train_loss_values = network.history['loss']
    train_accuracy_values = network.history['accuracy']
    val_loss_values = network.history['val_loss']
    val_accuracy_values = network.history['val_accuracy']

    plt.subplot(2,2,1)
    plt.plot(epochs, train_accuracy_values, label=f'LR = {lr}')
    plt.subplot(2,2,2)
    plt.plot(epochs, train_loss_values, label=f'LR = {lr}')
    plt.subplot(2,2,3)
    plt.plot(epochs, val_accuracy_values, label=f'LR = {lr}')
    plt.subplot(2,2,4)
    plt.plot(epochs, val_loss_values, label=f'LR = {lr}')
    
# Formatting for each subplot
ylabel = ['Training Accuracy', 'Training Loss', 'Validation Accuracy', 'Validation Loss']
for i in range(4):
    plt.subplot(2,2,i+1)
    plt.title(f'{ylabel[i]} over Epochs for Different Learning Rates')
    plt.xlabel('Epochs')
    plt.ylabel(ylabel[i])
    plt.legend()
    plt.grid(True)

plt.show()

I do not have your data, but the output should look like this:

Answered By - Mattravel