My recent data science and analytics projects:

Skin Cancer Image Detection:

Can deep learning help the everyday user know whether a mole is normal, or whether they should really get that checked out? This project leverages FastAI's ability to fine-tune pre-trained models to create an image classification model capable of distinguishing malignant skin lesions from benign.

The best model is an ensemble of 5 CNN learners fine-tuned on ResNet34. This model scored as AUROC of 0.85 on a validation set of 7,700 unseen images, indicating that it has an 85% chance of distinguishing between melanoma and benign skin lesions.

Global Solar Analysis Tableau Dashboard

A project which overlays global solar power generation capacity data on a background map which shows photovoltaic power potential. The dashboard is intended to be used as a tool for solar businesses assessing international market potential based on where solar power is and isn't being used along with which locations have the greatest potential.

And, the storydeck to explain can be found here:

Patient Diagnosis with Machine Learning

#A multiclass classification problem: given a patient's symptoms, can a machine learning model sucessfully give a diagnosis?

Python tools: Pandas, Scikit Learn, XGBoost, Random Forest, AdaBoost, SVM, GridSearchCV, RandomSearchCV, Matplotlib, Seaborn, Predictive Power Score, Feature Importances

Customer and Supply Chain Analytics for a Global Fashion Brand

Carefully constructed SQL queries to pull key insights out of 3 tables as part of a take-home assessment from a clothing retailer, along with answers to business scenarios accompanying each one.

Springboard Work:

Some selected work from my 550hr+ data science intensive course