Examine correlations of the new features

Now that we have our volume and datetime features, we want to check the correlations between our new features (stored in the new_features list) and the target (5d_close_future_pct) to see how strongly they are related. Recall pandas has the built-in .corr() method for DataFrames, and seaborn has a nice heatmap() function to show the correlations.

This exercise is part of the course

Machine Learning for Finance in Python

View Course

Exercise instructions

Extend our new_features variable to contain the weekdays' column names, such as weekday_1, by concatenating the weekday number with the 'weekday_' string.
Use Seaborn's heatmap to plot the correlations of new_features and the target, 5d_close_future_pct.

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

# Add the weekday labels to the new_features list
new_features.extend([____ + ____ for i in range(1, 5)])

# Plot the correlations between the new features and the targets
sns.____(lng_df[____ + ['5d_close_future_pct']].corr(), annot=True)
plt.yticks(rotation=0)  # ensure y-axis ticklabels are horizontal
plt.xticks(rotation=90)  # ensure x-axis ticklabels are vertical
plt.tight_layout()
plt.show()

Edit and Run Code