Hi, I'm Austin!
-
Vesuvius Challenge - Ink Detection EDA
- Context: This competition entailed developing an algorithm to detect ink from X-Ray CT scans of scrolls that were carbonized by a fire and had known characters and character locations. The EDA explores the 65 X-Ray CT slices by plotting layer-by-layer voxel intensity histograms as well as summary statistics at each layer; potential normalization and cross-validation strategies are recommended as a result. The notebook was cited by several top-10 performing teams and by an academic pre-print.
-
Vesuvius Challenge - Tensorflow implementation of tutorial
- The goal of this project was to re-implement the tutorial given in PyTorch with TensorFlow.
-
6.419x Data Analysis Written Reports
- Available upon request - made private to comply with academic integrity standards. The repo consists of 5 written reports over the 5 course Modules, most of which heavily utilized Python for the analysis. The Modules are,
- Module 1: Review: Statistics, Correlation, Regression, Gradient Descent
- Module 2: Genomics and High-Dimensional Data
- Module 3: Network Analysis
- Module 4: Time Series
- Module 5: Environmental Data and Gaussian Processes
- Available upon request - made private to comply with academic integrity standards. The repo consists of 5 written reports over the 5 course Modules, most of which heavily utilized Python for the analysis. The Modules are,
-
Probability Maximization and Financial Investment Analysis
- This project was carried out in the context of an MMORPG wherein I gave the community an analysis of the money/time investment likely required to obtain a particular rare item. In the context of the game, players are required to trade items to an npc for a chance of obtaining the desired quest items. One trade yields an item with probability 0.5% and the other yields its item with probability 0.4%. I assume that each trial is iid bernoulli and I progress to using the CDFs of the geometric distributions to perform the analysis.
-
- Simulated the outcome and estimated the distribution of a random variable,
$W = \lfloor \frac{Y_1 + Y_2}{2}M \rfloor$ , where$Y_1$ ,$Y_2 \stackrel{iid}{\sim} U\set{45, 150}$ and$M \sim U[4.65, 5.15]$ . The PMF of$\frac{Y_1 + Y_2}{2}$ is computed analytically prior to simulation to reduce the uncertainty in the estimated distribution.
- Simulated the outcome and estimated the distribution of a random variable,
