🏠 Keras Neural Network — House Price Prediction USING PYTHON CODE

🏠 Keras Neural Network — House Price Prediction

Learn Regression with a Simple Dataset + Copy & Execute Python

📚 1. What are we going to predict?

In this example, a neural network learns the relationship between house size and house price.

House Size (sq ft)	Price (₹ lakh)
500	20
600	24
700	28
800	32
900	36
1000	40
1100	44
1200	48

Goal: Train a neural network and estimate the price of a new house, such as a 750 sq ft house.

🧠 2. How the Neural Network Works

🏠 House Size
Input

→

🔵 Input Layer

→

🧠 Hidden Layer

→

💰 Price
Output

The model receives house size as input. During training it changes its internal weights until its predicted prices become close to the actual prices.

💻 3. Keras Python Code

import numpy as np import tensorflow as tf from tensorflow import keras from tensorflow.keras import layers # ----------------------------------- # 1. Training Data # ----------------------------------- X = np.array([ [500], [600], [700], [800], [900], [1000], [1100], [1200] ], dtype=float) y = np.array([ [20], [24], [28], [32], [36], [40], [44], [48] ], dtype=float) # ----------------------------------- # 2. Create Neural Network # ----------------------------------- model = keras.Sequential([ layers.Input(shape=(1,)), layers.Dense(10, activation="relu"), layers.Dense(1) ]) # ----------------------------------- # 3. Compile Model # ----------------------------------- model.compile( optimizer="adam", loss="mse", metrics=["mae"] ) # ----------------------------------- # 4. Train Model # ----------------------------------- model.fit( X, y, epochs=500, verbose=0 ) # ----------------------------------- # 5. Predict New House Price # ----------------------------------- new_house = np.array([[750]], dtype=float) prediction = model.predict( new_house, verbose=0 ) print("House Size:", new_house[0][0], "sq ft") print( "Predicted Price:", round(prediction[0][0], 2), "lakh" )

Output will appear here...

ℹ️ About browser execution:
The editor lets students copy and edit the real Keras/TensorFlow program. The browser execution button runs a lightweight Python demonstration of the same regression idea using Pyodide, because loading the complete TensorFlow/Keras runtime directly into a normal webpage is very large.

🔢 4. Mathematics Behind the Model

Neuron:

z = wx + b

ReLU:

ReLU(z) = max(0,z)

Here w represents the weight, x represents the input house size and b is the bias.

Mean Squared Error

MSE = (1/n) Σ(y − ŷ)²

The optimizer tries to reduce this error by adjusting the neural network's weights and biases.

⚙️ 5. Important Keras Commands

Command	Purpose
keras.Sequential()	Creates a sequential neural network
model.compile()	Defines optimizer, loss and metrics
model.fit()	Trains the neural network
model.predict()	Generates predictions for new data

🎯 6. What Should the Model Learn?

Looking at the training data, the relationship is approximately:

Price ≈ 0.04 × House Size

Therefore, for a 750 sq ft house:

0.04 × 750 = 30 lakh

A properly trained neural network should therefore produce a prediction close to ₹30 lakh.

🔄 7. Complete Learning Process

Dataset

→

Neural Network

→

Prediction

→

Loss

→

Weight Update

This process repeats for many epochs until the model learns a useful relationship between input and output.

🌐 Created by Bijan Krishna Paul

Keras Neural Network — House Price Regression Example