two_sigma_challenge

Two Sigma Data Clinic Data Test

Project Organization

├── LICENSE
├── Makefile           <- Makefile with commands like `make data` or `make train`
├── README.md          <- The top-level README for developers using this project.
├── data
│   ├── external       <- Data from third party sources.
│   ├── processed      <- The final, canonical data sets for modeling.
│   └── raw            <- The original, immutable data dump.
│
├── models             <- Trained and serialized models, model predictions, or model summaries
│
├── notebooks          <- Jupyter notebooks for exploration and development.
│
├── references         <- Data dictionaries, manuals, and all other explanatory materials.
│
├── reports            <- Generated analysis as HTML, PDF, LaTeX, etc.
│   └── figures        <- Generated graphics and figures to be used in reporting
│
├── src                <- Source code for use in this project.
│   ├── __init__.py    <- Makes src a Python module
│   │
│   ├── data           <- Scripts to download or generate data
│   │   └── make_dataset.py
│   │
│   ├── features       <- Scripts to turn raw data into features for modeling
│   │   └── build_features.py
│   │
│   ├── models         <- Scripts to train models and then use trained models to make
│   │   │                 predictions
│   │   ├── predict_model.py
│   │   └── train_model.py
│   │
│   └── visualization  <- Scripts to create exploratory and results oriented visualizations
        └── visualize.py

Reproducing Project

1. Run make data to get and clean the raw 2016 citibike trip data, and merge with cleaned NOAA data. Note the NOAA data was (unfortunately) downloaded through a GUI, and its provenance is given in "Order History | Climate Data Online (CDO) | National Climatic Data Center (NCDC).pdf".
2. Run make models to train, pickle, and save models. Note for this project, two models were quasi-optimized using grid search over a fixed validation set, specifically an elastic net (l1/l2 regularized linear model) and a Gradient Boosted Regression model (using least squares (l2) loss function). Note we say "quasi-optimized" because additional the hyperparameter searches could have been extended.
3. Run make model-performance-visuals to make learning curves for both models. Plots are saved in reports/figures/. TODO
4. Run make predictions to make final model, and get final test error.
5. Run make report to make final report. ....

---

Project based on the cookiecutter data science project template. #cookiecutterdatascience