Predicting Irrigation Need

This project focuses on building a deep learning model to predict health risks (such as diabetes) using a real-world medical dataset.

The project demonstrates the full deep learning pipeline, starting from data preprocessing to model optimization and evaluation. The dataset was cleaned by handling missing values and duplicates, followed by feature selection and normalization to prepare it for training.

A Feedforward Neural Network (FNN) was implemented using multiple hidden layers to capture complex patterns in the data. Different activation functions such as ReLU, Sigmoid, and Tanh were explored to enhance model performance.

To train the model effectively, Backpropagation and Gradient Descent techniques were applied, along with advanced optimizers like Adam and SGD. The model was trained over multiple epochs with careful monitoring of loss and performance.

To improve generalization and prevent overfitting, several regularization techniques were used, including Dropout, Early Stopping, and Model Checkpointing.

Hyperparameter tuning was performed using automated search methods (such as Random Search / Keras Tuner) to find the best combination of learning rate, number of neurons, and batch size.

The final model achieved improved accuracy and demonstrated the ability to generalize well on unseen data, making it suitable for real-world predictive tasks in healthcare analytics.