House Price Prediction

The Problem

Imagine that you are helping someone estimate the price of a house.

You know the size of several houses and you also know how much those houses were sold for.

Now someone shows you a new house and asks:

We don't know the exact price yet. But perhaps the previous houses can help us make a reasonable prediction.

Let's Look at the Data

Suppose we have information about three houses:

Look carefully at the numbers.

As the house size increases, the price also increases.

Can We Find a Pattern?

Instead of immediately using a Machine Learning algorithm, let's try to find the relationship ourselves.

Compare the first two houses.

So an additional 200 square feet is associated with an additional ₹10 lakh in this simple example.

How Much Does 1 Square Foot Add?

We can divide the price increase by the size increase.

₹0.05 lakh is ₹5,000.

We Can Write the Relationship as a Formula

We found that the price increases by ₹0.05 lakh for every additional square foot.

That gives us a simple relationship:

Let's check whether this formula works for our examples.

Now Predict the Price of a New House

Suppose a new house has a size of:

We can use the relationship we found:

We didn't know the actual price of the new house. We used the relationship found from the previous examples to make a prediction.

What Did We Actually Do?

We started with examples where both the house size and price were known.

But Real Predictions Aren't Perfect

Our example was deliberately simple.

In the real world, house prices do not depend only on size.

So the formula:

is useful for understanding the basic idea, but it would not be enough to accurately predict every real-world house price.