Javier Rodriguez ZaurininTowards Data Sciencepytorch-widedeep, deep learning for tabular data IV: Deep Learning vs LightGBM30 min read·Jun 13, 2021----
Javier Rodriguez ZaurininTowards Data Sciencepytorch-widedeep: deep learning for tabular dataA flexible package to combine tabular data with text and images17 min read·Feb 22, 2021--2--2
Javier Rodriguez ZaurininTowards Data ScienceRecoTour III: Variational Autoencoders for Collaborative Filtering with Mxnet and PytorchA deep dive into the use of Variational Autoencoders for Collaborative Filtering13 min read·Jun 19, 2020----
Javier Rodriguez ZaurininTowards Data SciencePredicting Amazon review scores using Hierarchical Attention Networks with PyTorch and Apache…Using the hierarchical structure of documents and MLP-based attention mechanisms to predict amazon reviews scores16 min read·Jan 4, 2020----
Javier Rodriguez ZaurininTowards Data ScienceLightGBM with the Focal Loss for imbalanced datasetsThe Focal loss (hereafter FL) was introduced by Tsung-Yi Lin et al., in their 2018 paper “Focal Loss for Dense Object Detection”[1]. It is…7 min read·Oct 6, 2019--3--3
Javier Rodriguez ZaurininTowards Data ScienceRecoTour II: neural recommendation algorithmsAdding Deep Learning-based recommendation algorithms to RecoTour, a tour through recommendation algorithms in python.15 min read·Sep 11, 2019--2--2
Javier Rodriguez ZaurininTowards Data SciencePutting ML in production II: logging and monitoringCombining MLflow and hyperparameter optimization tools to log and monitor all the useful data from ML pipelines15 min read·Mar 11, 2019--3--3
Javier Rodriguez ZaurininTowards Data SciencePutting ML in production I: using Apache Kafka in Python.Using a message broker to productionise algorithms in real time10 min read·Mar 5, 2019--1--1
Javier Rodriguez ZaurininDataDrivenInvestorRecoTour: a tour through recommendation algorithms in pythonA while ago a friend of mine asked me about approaches that would be useful when optimising GBMs. I had been asked this question a few…9 min read·Nov 29, 2018----