Data Science Handbook/

...

/

Presentation of Results

Presentation of Results

Get some hands-on experience on how to present and analyze results in Python.

We'll cover the following...

Results analysis

Visualization of results helps us understand and interpret the models’ performance. We can use Python resources, such as the seaborn and Matplotlib libraries, for this purpose. Scatterplots are good for visualizing relationships and checking for patterns. We can use them to see how our models predict the outcome based on a single variable of the data. Additionally, we’ll create a best-fit line on each scatterplot, with the x-axis representing the actual tip amount and the y-axis representing the predicted amounts.

We’ll analyze the results by generating scatterplots that utilize two variables from the tips dataset: the time variable and the sex variable. Additionally, we’ll produce a histogram to directly compare predictions with the actual data.

Remember: When we execute any of the code below again, we might have different results because the models will be trained again from scratch.

Based on the time variable

Because this column only had Lunch or Dinner as options, we’ll see how the predicted values align with the original values for each option.

Press + to interact
#Import the necessary libraries
import matplotlib.pyplot as plt
import seaborn as sns
# Flatten the neural network predictions to make them compatible with the plotting function
nn_predictions = nn_predictions.flatten()
time_mapping = {1: 'Lunch', 0: 'Dinner'}
# Combine the actual and predicted tip values with the 'time' feature for each model
lr_df = pd.DataFrame({'Actual Tip': y_test, 'Predicted Tip (LR)': lr_predictions, 'Time': X_test['time_Lunch'].map(time_mapping)})
rf_df = pd.DataFrame({'Actual Tip': y_test, 'Predicted Tip (RF)': rf_predictions, 'Time': X_test['time_Lunch'].map(time_mapping)})
nn_df = pd.DataFrame({'Actual Tip': y_test, 'Predicted Tip (NN)': nn_predictions, 'Time': X_test['time_Lunch'].map(time_mapping)})
# Create separate scatterplots for each model
sns.lmplot(x='Actual Tip', y='Predicted Tip (LR)', data=lr_df, hue='Time', markers=['o', 'x'])
plt.title('Linear Regression Predictions')
plt.xlabel('Actual Tip')
plt.ylabel('Predicted Tip')
plt.savefig("output/plot2.png", bbox_inches='tight')
sns.lmplot(x='Actual Tip', y='Predicted Tip (RF)', data=rf_df, hue='Time', markers=['o', 'x'])
plt.title('Random Forest Predictions')
plt.xlabel('Actual Tip')
plt.ylabel('Predicted Tip')
plt.savefig("output/plot1.png", bbox_inches='tight')
sns.lmplot(x='Actual Tip', y='Predicted Tip (NN)', data=nn_df, hue='Time', markers=['o', 'x'])
plt.title('Neural Network Predictions')
plt.xlabel('Actual Tip')
plt.ylabel('Predicted Tip')
plt.savefig("output/plot.png", bbox_inches='tight')

Code explanation

  • Lines 2–3: We import the necessary libraries.

  • Line 6: We flatten the neural network predictions to ensure compatibility with the plotting function.

  • Line 8: We create a mapping dictionary to convert binary values to Lunch and Dinner for the time variable.

  • Lines 11–13: We create DataFrames combining the actual and predicted tip values for ...