Nathan Arias
Hello I'm Nathan! I am a molecular biologist turned data scientist with a concentration in NLP and data visualization. As a data scientist at Amgen and an imminent UC Berkeley Data Science Master’s graduate, I am enthusiastic about applying my diverse expertise to innovate across various tech sectors, always ready for new challenges and breakthroughs.
Los Angeles, CA
Projects
Citibike Station Analysis
In this project, we analyzed ten years of CitiBike customer data, mapping journey start and end points to produce a neighborhood-based choropleth. This visual representation of user patterns informed our recommendations for new station placements.
HAM10000 Skin Lesion Classifier
Developed a mixed model using a Convolutional Neural Network (CNN) with the HAM10000 dataset, achieving 81% accuracy in skin lesion classification. The project entailed rigorous data augmentation and fine-tuning of neural network layers to optimize performance and accuracy
Optimized Pairing Analysis: Enhancing E-Commerce Strategies with Graph-Based Data Science
This project involved analyzing the Northwind database to identify products often purchased together, using graph-based Bayesian analysis and community detection algorithms such as Louvain Modularity. The insights were applied to enhance both online product recommendations and warehouse storage layouts.
Precision Confluency Analysis in Live/Dead Stain Assays
This application employs dual-channel live/dead staining to enhance microcarrier segmentation through the exploitation of autofluorescence properties. It adeptly performs binary inversion and subtraction on the segmented images to remove non-adherent floaters. The core functionality lies in its sophisticated algorithm, which calculates confluency by quantifying the ratio of the microcarriers’ area to the live signal’s area, thereby offering a precise and reliable metric for cell culture analysis in biotechnological research.
