Name: Edward D. Ramirez
Type: User
Company: Rutgers University
Bio: Physics PhD candidate at Rutgers University performing data-driven studies of dark matter. Interested in quantitative finance and data science projects.
Location: Highland Park, NJ
Blog: https://edwarddramirez.github.io/
Edward D. Ramirez's Projects
This repo provides code for detecting point sources on a sphere (i.e., the sky) by applying the continuous wavelet transform on raw count data. We use this tool to detect faint gamma-ray point sources from Fermi-LAT. See my skysearch repo for a pipeline.
Summary notebooks using derivative gaussian processes with tinygp. We implement a 2D derivative gaussian process and successfully use derivatives to regularize SVI fits with a gaussian process model..
Code used to constrain dark matter substructure in the solar neighborhood with Gaia eDR3 wide binaries.
Github Pages template for academic personal websites, forked from mmistakes/minimal-mistakes
A study of the possible relationship between company lobbying activity and company stock price movement.
Summary notebook implementing Bayesian Model Averaging with numpyro.
Simple 2D continuous wavelet transform implementation in Python
Pipeline for finding point sources in Fermi-LAT data with the continuous wavelet transform. For a tutorial that describes how this code works, see the allsky-point-source-detection repo.
Bayesian inference using sparse gaussian processes from tinygp. Examples include 1D and 2D implementation.
Novel technique to fit a target distribution with a class of distributions using SVI (via NumPyro). Unlike standard SVI, our "data" is a distribution rather than a finite collection of samples.
Forecast NYC taxi activity with deep learning. We compare the performances of models based on MLPs, RNNs, LSTMs, GNNs, and ARIMAX. Additionally, our code provides users with an easy-to-use pipeline for producing custom time series datasets of taxi activity from publicly available NYC TLC data.
1D Density Estimation with the Haar Wavelet. Application is to separate point sources from a background distribution.