- Jupyter Notebook:
pip install jupyter
- Jupyter Lab:
pip install jupyterlab
- nteract
- google colab
These two sites are great resources
- Machine Learning Mastery: Provides plenty of tutorials with python code that you can experiment with.
- Chris Albon: Great visualizations of how things work.
Thomas Long to present back propagation and gradient descent
The next meeting is going to consist of the following parts.
- Education piece about a TBA machine learning algorithm – may be fulfilled by following talk
- Tooling/education about Scikit learn metric tools
- Application/education of reinforcement learning to blackjack – how to create a table that tells you what the best move is.
- Michael Smith will share his results from the sports dataset.
We would like 1-2 people to also either present their results and experiences with the sports data set or any other problem they are trying to apply machine learning to.
https://github.com/tdpetrou/Learn-Pandas This is a great resource for learning pandas.
For this meeting the education component is going to consist of:
- Matt discussing supervised learning and describe both classification and regression problems.
- Jason covering pandas. Go Pandas!!
For the application part of the meeting we want to see some members work based on the added sports data that is described below along with some tasks associated with it.
There is now a data folder in this project. The first dataset we are going to work with are some NFL sports data that I retrieved from https://www.kaggle.com/tobycrabtree/nfl-scores-and-betting-data/home. There are three files that may need to be joined to make the data more useful.
There are two supervised tasks for this dataset.
- Predict the spread.
- Predict the winner.
There are two sub-tasks that we could quickly think of for this data.
- Identify feature importance for either of the supervised tasks.
- Enrich the data with rosters at each game or anything else that seems reasonable.
You can access the data by checking out the project or by downloading it from google colab. You could then download a copy of the notebook and push that to the project, send it to Jason Mauzey, or remote in to your computer to show what you have done. Any work is good work. We look forward to seeing some good efforts.
If you don't want to checkout the project you can use these links to get the data, or use the above kaggle link. https://raw.githubusercontent.com/mauzeyj/machine_learning_club/master/Data/nfl_teams.csv https://raw.githubusercontent.com/mauzeyj/machine_learning_club/master/Data/nfl_stadiums.csv https://raw.githubusercontent.com/mauzeyj/machine_learning_club/master/Data/spreadspoke_scores.csv
Importing the data is very easy with pandas. pd.read_csv('https://raw.githubusercontent.com/mauzeyj/machine_learning_club/master/Data/spreadspoke_scores.csv')
I haven't started working with these data sets yet so I can't vouch for them yet. Lets all have a great time!!