Activity 5
AI to Predict House Prices

How does an AI predict house prices?

AIs learn from historical data. An AI that has analyzed the historical prices of a neighborhood-based on information such as square footage, age etc can be used to predict the price of any house in the neighborhood based on the same characteristics/features such as square footage, age etc.

Lets collect the data needed to build the AI

First, we need to choose a neighborhood and collect data about it. You can do this from any public dataset.

Here are a couple of options for collecting house prices data.

1. Download data from a public repository: (a) Option 1 (b) Option 2

2. Download data from Redfin for a neighborhood of your choice. Video below shows how this can be done.

Let's build an AI to predict house prices

  • We will use the dataset we collected in the previous step to build this AI.

  • When we train the AI to predict the house prices, it might do a good job at predicting a few better than others. We will use an error metric called RMSE to measure the performance

  • A low value of RMSE indicates aa good AI. A high value indicates a high error and hence a low performing AI.

  • An RMSE of 0 implies the AI predicts the price perfectly each time. Typically, this does not happen in the real world.

Let build our AI

​​How to start using Navigator

1. Go to

2. Follow the instructions in the video below to build the AI

Observe and Discuss


1. What is the Error of the AI?

2. Try a few houses from the dataset and see how close the AI predicts the price 

3. What can you do to improve the performance of this AI? Think about any additional information you can provide to help it make better predictions