From the course: Machine Learning and AI Foundations: Value Estimations
Unlock the full course today
Join today to access over 22,600 courses taught by industry experts or purchase this course individually.
Use as few features as possible: The curse of dimensionality
From the course: Machine Learning and AI Foundations: Value Estimations
Use as few features as possible: The curse of dimensionality
- [Instructor] When building a machine learning model it might seem like a good idea to include as many features in the model as possible. More data is always better, right? Well, it turns out that's not always true. When we talk about more data in our data set, we can mean two different things. First, we could be talking about more rows of data in the data set. For our housing data set, having more records of more homes is pretty much always a good thing. But we can also mean having more features or columns in the data. Having more features is helpful until a certain point. But you eventually hit a wall where more features hurts accuracy. The reason is called the curse of dimensionality. The curse of dimensionality says that as we add more features to our model the number of rows of training data that we need to build a good model tends to grow exponentially. Let's look at an example. Let's pretend that our house price predictor used only one feature, the house size and square feet…
Practice while you learn with exercise files
Download the files the instructor uses to teach the course. Follow along and learn by watching, listening and practicing.