projects.
Predictive Justice & explainability
Prediction on Italian IMU tax litigations using two neural network, Gru & LSTM, custom Attention layer for each. Focus on explainabilty through visualization of attention layer weights on text, LIME and N-GRAM analysis
Python
Tensorflow
LIME
PM2.5 Universal Kriging
Analysis of a public dataset from U.S.A - EPA. Focus on pm2.5, prediction using kriging interpolation.
R
Pharma sales time series
Classic and modern time series forecasting, different prediction and accurate analysis.
R
ARIMA
Wisconsin breast cancer prediction
Dataset with 29 features about brest cancer. Principal component analysis followed by a logistic regression; decision tree alghoritm for a different perspective.
R
pca
rpart
