Jupyter Notebooks

Collection of Jupiter Notebooks I have made. This could be anything from Code Alongs to Model Testing

1. Deep Reinforcement Learning - Followed a Tutorial on how to build a Deep Reinforcement Learning Model using Deep Q Learning. This was done on the Cartpole Game.

2. Text Classification - I built a bag of words deep learning model that handles text classification with 80%+ accuracy.

3. Testing AutoViz - I tested out a new Python package, AutoViz, against a couple of different datasets. Results were interesting.

4. CycleGAN Explained - My attempt at explaining the CycleGAN paper in my own words as simply as possible.

5. Currency Predictor - I found a sample assignment online that looked interesting. Involves exploring JPY/USD price pair over time in an attempt to find a trading pattern.

6. Currency Predictor PT2 - Here I took the information learned in PT1 and used that to clean up my code and simulate some strategies

7. Lawmaker Tracking Data Preprocessing - There is a movement online of traders tracking the trades that lawmakers make in order to attempt to copy inside trading. My strategy is a little bit different. I am not sure whether or not lawmakers are inside trading, but I am certain that there is a movement of people that are making trades based on that information. My theory is there will be a jump in each stock as the time of discloser due to this wave of people buying. Potentially selling aswell. If I can find that pattern, there is something I can consistently trade on.

8. Lawmaker Tracking Exploratory Analysis - Continuation of Lawmaker Tracking Data Preprocessing

9. BASALT 2022 Item Classifier - Helped my BASALT 2022 Competition submission by training a model to recognize the items in a minecraft inventory using solely a jpg of the screen.

10. BASALT 2022 Item Quantity Classifier - Helped my BASALT 2022 Competition submission by training a model to recognize each item's quantity in a minecraft inventory using solely a jpg of the screen.

11. BASALT 2022 TF-IDF MineWiki Search Engine - Helped my BASALT 2022 Competition submission by easily allowing me and my team to get relevent information out of the MineDojo dataset. Also used SparkML to keep it quick and efficient.

12. BASALT 2022 Move Classifier Data Exploration - Helped my BASALT 2022 Competition submission by allowing me and my team to sample and review training data for our MoveClassifier model

13. BASALT 2022 Data Generator Class Setup - Helped my BASALT 2022 Competition submission by setting up and testing my custom generators StartImageGenerator and EndImageGenerator.

14. BASALT 2022 Create Search Engine for MineDojo Youtube Dataset - A notebook demonstration of how I use TF-IDF to search through transcripts of 70k gameplay Youtube Videos distributed by MineDojo. After the success on the MineWiki search, I brought that to this problem. Doesn't work great because the words said are not as consistent in a transcript as a Wiki page. It still uses PySpark for all the data preprocessing, transformation and TF-IDF work which makes it really fast. After this I need to look into some other options for gathering information from transcripts.

15. BASALT 2022 Explore Depth and ColorMaps - A notebook visualization of a Malmo dataset I generated for training data to feed into BlockSegmentator.py and DepthEstimator.py. Lots of visuals, tables, and pretty pictures.