Boston-Housing-Prediction-PyTorch
This neural network predicts Boston house prices by using a hidden layer to learn complex, non-linear patterns in the data. The model uses data cleaning and feature scaling, ensuring the math remains stable. By using a Dropout layer to prevent the model from simply memorizing the training set, it is able to generalize the data.
Dataset used : https://www.kaggle.com/datasets/altavish/boston-housing-dataset
Feature Description :
CRIM: The per capita crime rate in the town.
ZN: The proportion of residential land zoned for large lots (over 25,000 sq. ft.).
INDUS: The proportion of non-retail (industrial) business acres per town.
CHAS: A "dummy" variable that is 1 if the property is on the Charles River and 0 if not.
NOX: The concentration of nitric oxides in the air (a measure of pollution).
RM: The average number of rooms per house.
AGE: The proportion of owner-occupied units built before 1940.
DIS: The weighted distance to five major Boston employment centers.
RAD: An index showing how easy it is to access major radial highways.
TAX: The property tax rate per $10,000 of value.
PTRATIO: The pupil-teacher ratio (number of students per teacher) in town schools.
B: A calculation based on the proportion of Black residents in the town.
LSTAT: The percentage of the population considered "lower status" based on socio-economics.
Target variable:
MEDV: The median value of owner-occupied homes in $1000 s (your Target variable).