Boston Report
In this project, I embarked on a journey to predict Boston house prices
using machine learning and deep learning techniques. The central focus
of the project was to construct a predictive model that outperforms
traditional linear regression. By harnessing the power of neural
networks.
Highlights:
-
Baseline Linear Regression: I initiated the project by
establishing a baseline using a linear regression model from the
scikit-learn library. This served as a foundational benchmark against
which the neural network's performance could be evaluated.
-
Neural Network Implementation: The project's centerpiece was
the development of a neural network model tailored specifically for
predicting Boston house prices. This neural network brought the
advantages of non-linearity and complex feature interactions, offering
a superior modeling approach.
-
Feature Engineering: To enhance model performance, I conducted
thorough feature engineering, selecting and transforming relevant
features to ensure optimal predictive accuracy.
-
Hyperparameter Tuning: I fine-tuned the neural network's
hyperparameters to achieve the best possible performance, iterating
through various configurations to maximize predictive power.
Outcomes:
Reveals compelling insights by comparing a simple linear regression
model to a neural network using a straightforward dataset.
Full report in PDF