Description of this portfolio

This portfolio contains my notebooks about machine and deep learning algorithms applied on time series data. This data was downloaded from different sources and it's freely availble. If you can't see my notebooks here, you can use nbviewer. The notebooks are described below:

1. XGBoost for COVID-19 time series data

The purpose here was to do baselines using XGBoost. The code was written using R and its ecosystem (e.g. tidyverse). The data it's about COVID-19 and comes from several countries. After training, i computed feature importance to give explainability to the model.

2. XBNet vs XGBoost Predictions on COVID-19 data

XBNet (Extremely Boosted Neural Network) it's another neural network specially designed for tabular data. As the name allows to infer, it uses an approach similar to XGBoost. Here, the predictions made with XBNet are better than XGBoost.

3. GBM + NNs, TabNet, XGboost and more for COVID-19 data

This is a kind of benchmark of several models trying to find the best one to model COVID-19 data from India. It's interesting how the TensorFlow ecosystem allows to connect the outputs of a neural network to a gradient boosting machine using TensorFlow Decision Forest library. I also used TabNet, an amazing neural network architecture which has a realy good performance on gene expression data.

4. A Predictor for Batery's data

As freelancer I'm not always allowed to share code, however, I don't have any problem with this piece of code. Here, again a benchmark is showed, with interesting models like Temporal Convolutional Netowork (TCN) and more.

Contact: [email protected]