Gaspare Mattarella's repositories
llm4sql
Bridging the Gap Between AI and SQL Databases
OpenDevin
🐚 OpenDevin: Code Less, Make More
Spotify-Playlists-Downloader
This simple script is meant to go through all your personal account playlists (both strictly yours and followed), and download every single track in it.
Students_Performance-Analysis
In the first part of this paper we are going to perform a regression analysis on a dataset concerning students performances in secondary school in Portugal. Our goal is to find the variables that most explain the variances, understand how and possibly why this would be the case. To achieve this goal we will use with different models starting from the basic linear regression and going on selecting the best features with a stepwise selection model, a LASSO and finally a Robust regresssion. Then we will try to obtain additional informative power thanks to two Tree Based models. In the second part of the assignment we will instead see how a simple K means algorithm can well divide the dataset in two clusters representing good and bad performative students.
Assessing-the-Effectiveness-of-Unconventional-Monetary-Policy-in-the-Euro-Area
This paper assesses the macroeconomic effects of Unconventional Monetary Policy
stat_rethinking_2022
Statistical Rethinking course winter 2022
Open-source-data-hub-for-italian-geospatial-information
Open source data hub for italian geospatial information
Suicides_Shiny_App_Dashboard
This dashboard is intended to provide some general insight on the delicate matter of suicides all over the world based on a dataset with data that range from 1985 to 2015 for every country accounting for sex, age and economic variables such as per capita GDP.
A_Panel_Data_Analysis_to_Forecast_US_Presidential_Elections
In this paper I will initially retrace the path marked by Ray C. Fair with his long lasting series of presidential elections forecasts exploiting the same variables he uses but enriching the model with panel data. Exploiting the Fixed Effects estimation I will then add new variables that, according to our intuition, could lead to an overall improvement of the model and test for them applying the LASSO algorithm for model selection. I will finally infer the results and explore the possible challenges in disentangling causality from correlation
Marshall_and_Labor_Demand_in_Russia_Data_Analysis
Inference of the Russian labour demand function for the years 1996 and 1997 estimating and testing hypothesis on OLS, Instrumental Variables, FE, RE and Pooled Panel Data Models assessing how to better build a model for our data.
A_Brief_Time_Series_Analysis_of_German_Bund_Term_Structure_of_Interest_Rate
Testing the hyphotesis of cointegration of two term structures through Dickey-Fuller tests and Engle-Granger causality. Finally I exploit the VECM to infer the model and through the Cholesky decomposition I analyze SIRF and FEVD
An_NLP_algorithm_for_interactive_prediction
I designed a naive shiny web application which is intended to take a string of words and predict the next possible word based on the probability of occurrence exploiting Markov chains.