This is a collection of ML, Data Science and AI projects I have done for academic and self-learning purposes. In case you need more info about them:
This is my master thesis project on information extraction using LLMs and RAG. In this project I am building a web application where you can upload insurance documents. Once you have done that, you can use the chatbot function to ask questions to the LLM about the submitted documents. N.B. This is a simplified version that I developed after submitting my Master's thesis project and that I am using as a demo because the original material, along with all the testing and fine-tuning material, is copyrighted by the company.
This project focuses on developing a meta-learning approach to optimize hyperparameters for the XGBoost algorithm. By leveraging a large dataset of XGBoost configurations evaluated across numerous classification tasks, the goal is to predict the best configurations for new classification problems. The project includes experimenting with various performance predictors, evaluating useful meta-features, and identifying the most influential hyperparameters to enhance the default configuration.
Bachelor Thesis: Customer Lifetime Value (CLV) forecasting through the use of Recency Frequency and Monetary (RFM) variables.
This projects aims at forecasting the CLV using RFM variables and other simple variables obtained from Customer Relationship Management (CRM) data. The models used range from the linear model to the Rando Forest. In particular, I used a simple linear model as a baseline and then tried to modify the target variable (log-lin and quantile transformation) to get more stable results. I also tried to use RandomForest errors to try to detect unseen patterns in them. For the RandomForest model, I also tried to modify the target variable (log transformation). Finally, I tested different data transformations to see how the models behave. The comments are in Italian.