Sofia Pasquini

Sofia Pasquini

MSc. Astrophysics Student at Western University Researching variability in large carbon molecules in space using JWST data and machine learning techniques.

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

Posts

Future Blog Post

less than 1 minute read

Published:

This post will show up by default. To disable scheduling of future posts, edit config.yml and set future: false.

Blog Post number 4

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 3

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 2

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 1

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

portfolio

Neural Network Classification

A deep learning, neural network classifier constructed using the Keras framework. Model hyperparameters were trained/optimized using a 5-fold cross-validation grid search appraoch in order to be able to predict whether an individual has diabetes or not using the Pima Indian Diabetes Database. Leveraged the visualization capabilities of TensorBoard in order to visualize the training process. Click here to see this repository.

Sequence-to-Sequence Encoder-Decoder RNN

I explore building a sequence-to-sequence NLP solution using an Encoder-Decoder model built with the Keras framework that could predict the expanded form of a polynomial based on the input factored form. Click here to see this repository.

Probing Evolution of Carbonaceous Molecules Using Clustering Analyses

An exploration of the efficacy of clustering analyses on astronomical data (with observations from Spitzer and James Webb Space Telescopes) I perform for my M.S.c research. I hope to understand the mechnism of the evolution of stable, carbonaceous molecules in high-energy star-forming regions by analyzing key stratifications in emission properties from these molecules on various spatial and wavelength scales. To see this repository Click here.

Ensemble Regression - Housing Price Predictor

I experiment with various advanced ensemble, regression techniques such as a Bagging Regressor, a Random Forest Regressor, and a Gradient Boosted Regressor in order to predict the selling price of a house based on property description features. Click here to view this repository.

Numerical Simulation to Solve Lane-Emden Equation.

I use a system of ordinary differential equations and a second-order Runge-Kutta integration scheme in order to numerically simulate the physical equation of state describing the interior structure of a stellar body. I solve the Lane-Emden equation for the solutions of order n=1.5, 3, 3.5, though this value can be altered to higher dimension for further exploration. Click here to view this repository.

NLP Application: Spam vs Ham

I explore using Natural Language Processing techniques (punctuation removal, tokenization, stopword removal, lemmatization, and vectorization) to build a sample spam-trap to distinguish between ‘spam’ or ‘ham’ text messages. This classifier (I compare the performance of a Random Forest and Gradient Boosted Classifier in this context, and for my purposes find the Random Forest algorithm to be better suited. Click here to see this repository.

Python ETL Package- SpecFetch23

This repository contains a published Python package of my own creation which acts as an ETL pipeline that cleans and serves energy distribution data for any given object from the Sloan Digital Sky Survey database queried by the user using the object’s Celestial Coordinates (right ascension and declination). Though this package uses the celestial coordinates and SDSS database to query for data, these parameters can be easily varied in future releases. Click here to see this repository.

Consumer Activity Clustering

I use clustering (K-Means) in order to segment a consumer demographic based on Recency, Frequency and Monetary Value in order to improve quanitative marketing strategies of a business and to develop commercial strategies for a given company for which this data is available. Click here to see this repository.

Sentiment Analyis Application: IMDB Reviews

I use dimensionality reduction techniques (PCA, Autoencoding) in order to analyse and predict the sentiment in a set of reviews from the Internet Movie Database (sentiments originally labelled by humans). I explore and compare the predictive performances of dimensionally-reduced input using Partial Component Analysis and Autoencoding. Click here to see this repository.

Multi-modal CNN

I explore multi-modal learning using a Convolutional Neural Network to predict housing prices based on numeric property descriptors as well as images of the interior/exterior of the house. I employ a grid search functionality in order to tune and assess effects of various hyperparameter combinations. Click here to see this repository.

publications

talks

teaching

Teaching experiences

Undergraduate course, Western University, Earth Sciences, 2020

ES2213: The Dyanamic Earth

  • Duties for this course included:
    • Creating assignment, midterm, and final exam questions.
    • Holding office hours to answer questions.
    • Providing tutorials in MS Excel and MS Word for geophysical data processing and analysis.
    • leading weekly lab sessions.
    • Marking assignments and exams.

Teaching experiences

Undergraduate course, Western University, Earth Sciences, 2021

ES2232: Exploring the Planets

  • Duties for this course included:
    • Marking assignments, tests, quizzes and labs.
    • Correspondence with students.