Every feature is created different from others. Consider the house pricing problem again. Let's look at some of the features of a house:

• House size
• Lot size
• Number of rooms
• Number of bathrooms
• Type of parking garage (carport versus covered)
• School district
• Number of dogs barking in the house
• Number of birds chirping in backyard trees

The last two features may look ridiculous to you, and you might wonder what that has got to do with the house price, and you are right. At the same time, if these features are given to the machine learning algorithm, it may come up with a non-zero correlation to your surprise. This is where domain knowledge comes into the picture. It could be a data scientist wearing a domain expert hat or a domain expert wearing a data scientist hat, who would most likely do this job.