Hi, I'm Matthew

This brain tumor prediction model harnesses a fine-tuned CNN architecture, leveraging extensive data augmentation and rigorous preprocessing to achieve over 96% accuracy. Outperforming expert radiologists, it integrates advanced deep learning techniques with meticulous optimization to reliably detect and classify tumors. Designed for precision and efficiency, this model serves as a powerful diagnostic tool, supporting early detection and informed medical decisions in critical healthcare scenarios.

The Depression Predicting Model is a machine learning model that predicts whether a person has depression. It is implemented with scikit-learn and designed to make predictions based on various input features, carefully adjusted and refined. It employs robust preprocessing, data analysis, PCA, and feature engineering, combined with an SVM model, due to the high dimensionality and nature of the data. This is broad yet straightforward approach makes it well-suited for tabular datasets with moderate complexity, ensuring slight interpretability while leveraging the power of SVM, which is well-suited for this binary classification task.

The Life Expectancy Model is a deep learning model implemented in TensorFlow, designed to predict life expectancy based on various input features. The data was carefully adjusted and refined for the problem before the model was created. It employs a relatively shallow architecture with a few dense layers and uses the Adam optimizer for efficient, adaptive gradient descent, minimizing a linear regression loss to closely align predictions with actual values. The model achieves excellent regression performance, with an average prediction error of approximately ±1.2 years, and with good confidence intervals, which is a competetive result.

This is the prepared dataset to predict whether the person died or survived the Titanic disaster for a Kaggle competition. The code is highly flexible and includes hints at each step, allowing you to make modifications as needed. Additionally, it is provided as a notebook, so you can experiment with the code before using the dataset for ML/DL tasks. It covers data engineering, preprocessing, and adjustments tailored to ML and DL architectures, complete with detailed annotations.