# Netflix Federated Learning

Hey friends! Open invite to do some Federated Learning (or related PETs/data science) on my private Netflix data.

## Prerequisites
1. Install SyftBox by running:
   ```bash
   curl -LsSf https://syftbox.openmined.org/install.sh | sh
   ```
2. Follow the prompts to set up your sync directory and email
3. Need help? Check out the [SyftBox documentation](https://syftbox-documentation.openmined.org/) or join us on [OpenMined Slack](https://slack.openmined.org)

## Finding This Project
After installing SyftBox, you can find this project in your sync directory at:
```
<your-syftbox-directory>/datasites/andrew@openmined.org/public/netflix_fl/
```

You can also view this project online at:
https://syftbox.openmined.org/datasites/andrew@openmined.org/public/netflix_fl/

## Quick Start
1. Copy "example_job" to a new folder with your name and date (e.g., "suzy_prakash_job_nov_21")
2. Edit `job.py` to specify your computation
3. Edit `README.md` in your job folder to include:
   - Your name and contact details
   - A description of what your computation does
   - How to reach you (e.g., Slack, email)
4. Test locally with `./run.sh` (uses mock data)
5. Submit by moving your job to "submit_jobs_here" folder

## Project Structure
.
├── example_job/               # Template job to copy
├── mock_dataset_location/     # Test data for local development
├── submit_jobs_here/          # Drop your completed jobs here
└── completed_jobs_here/       # Find your processed results here

## Development
- Use `./run.sh` to test with mock data
- Use `./run.sh --real` to test with real data (only works for me)
- The mock dataset follows the same schema as the real Netflix viewing history (here's a [tutorial](https://www.slashgear.com/894320/how-to-see-and-download-your-entire-netflix-watch-history/) for how to download yours if you want to test locally or enable other people to train on your data!)

## Data Format
The Netflix viewing history CSV contains:
```csv
Title,Date
"Show Name: Season X: Episode Name","MM/DD/YY"
```

## Guidelines
- Please ensure your job runs successfully with mock data before submitting
- Keep computations reasonable (no training for days)
- Results will appear in "completed_jobs_here" after I run them

## Future Improvements
- I'll make it so that it sends you an email when your job is ready (when it arrives 
in "completed_jobs_here")

## Questions?
Feel free to reach out on slack (slack.openmined.org my user name is @trask) if you need help or have questions!

I look forward to seeing your projects!