Skip to content

Latest commit

 

History

History
128 lines (83 loc) · 4.73 KB

README.md

File metadata and controls

128 lines (83 loc) · 4.73 KB
Raw

Figment

Collection of projects for data science, attempts at kaggle competitions, and learning new languages/techniques.

Getting Started

In the different directories, I have Jupyter notebooks that are written in the respective languages for each project. This is structured so that I can organize different notebooks and test the relevant libraries and techniques.

Not all 'projects' will be written in multiple languages, but I will try to do that to have a nice comparison.

Python

The */python/ notebooks are designed to work on Google Colab.
Where possible, I'm trying to use Python3 because I have little experience with it compared to Python2.

My previous experience with Python has mostly involved high-energy physics analysis (software frameworks, plotting, and neural networks). Here I am developing some solutions to kaggle competitions and trying out new libraries and techniques.

Julia

I don't have any prior experience with Julia, so this is also a way for me to experiment with the language and develop some computer science skills. The */julia/ notebooks are designed to work on JuliaBox.

NB: There may be unintended incompatibilities based on what packages JuliaBox lets you download versus what I have available locally.

R

I also don't have any prior experience with R, so this is another way for me to practice with this language.
The */r/ notebooks are only tested locally, for now, but I will be investigating open source platforms in the future (Google Colab, Azure Notebooks, etc.).

NB: I downloaded R onto my macbook from here, version 3.5.1. And I used this reference for installing a jupyter kernel.

Projects

Housing Prices

The kaggle housing prices challenge is being explored with a different set of notebooks in both python and julia.

Notebooks

Here is a table with a brief description of each notebook contained in the repository.

Notebook About
python/0-housing_prices.ipynb Very basic and simple approach using one variable.
python/1-housing_prices.ipynb Basic multivariate / ML techniques (including BDT with XGBoost)
python/2-housing_prices.ipynb Basic deep learning approach with keras+tensorflow
python/3-housing_prices.ipynb Basic deep learning approach with pytorch
julia/housing_prices.ipynb Learning Julia with basic deep learning approach with Flux

Fantasy Football

As a personal exercise in data science, I'm attempting to make some (basic) fantasy football predictions using information from nflgame.

For now this is only a python project so I can practice SQL (via sqlite).

Notebooks

Here is a table with a brief description of each notebook contained in the repository.

Notebook About
python/0-fantasy-football.ipynb Read NFL data from nflgame and store relevant information in sql database.
python/1-fantasy-football.ipynb Very basic and simple approach for making predictions for one player.
python/2-fantasy-football.ipynb Preliminary attempt to use play-level information for making game-level predictions.

Personalized Medicine

To practice with natural language processing and using R, this project will explore the Kaggle challenge Personalized Medicine: Redefining Cancer Treatment.
NB: The training dataset was too large for GitHub, and will need to be downloaded separately from the Kaggle site.

Notebooks

Here is a table with a brief description of each notebook contained in the repository.

Notebook About
r/0-medicine.ipynb Exploring the dataset and getting familiar with R.

Misc.

Some references for performing the work in this repository:

Questions, Comments, and Concerns

Please post an issue or submit a PR.