https://github.com/aybchan/uber-time-series

Bayesian time series prediction
https://github.com/aybchan/uber-time-series

deep-learning probabilistic-forecasting pytorch time-series

Last synced: over 1 year ago
JSON representation

Bayesian time series prediction

• Host: GitHub
• URL: https://github.com/aybchan/uber-time-series
• Owner: aybchan
• Created: 2020-05-09T12:04:19.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2020-10-12T17:16:38.000Z (over 5 years ago)
• Last Synced: 2023-03-01T03:55:57.124Z (over 3 years ago)
• Topics: deep-learning, probabilistic-forecasting, pytorch, time-series
• Language: Jupyter Notebook
• Homepage:
• Size: 18.3 MB
• Stars: 41
• Watchers: 3
• Forks: 14
• Open Issues: 4
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Bayesian time series prediction

![Time series predictions with test set data](docs/predictions.gif)

End to end implementation of paper [Deep and Confident Prediction for Time Series at Uber](https://arxiv.org/abs/1709.01907) in PyTorch. We use the [Metro Interstate Traffic Volume](https://archive.ics.uci.edu/ml/datasets/Metro+Interstate+Traffic+Volume) multivariate time series dataset for training and eventually predicting traffic volume.

We make use of the implementation of [variational dropout](https://arxiv.org/abs/1512.05287) from [keitakurita/Better_LSTM_PyTorch](https://github.com/keitakurita/Better_LSTM_PyTorch) for the LSTM layers with dropout.

## Workflow



Code is prototyped in the [`notebooks`](notebooks) before transfer into cleaned up Python scripts for reuse.

- [`notebooks/01_dataset_creation.ipynb`](notebooks/01_dataset_creation.ipynb)

  - Download dataset from UCI repository, preprocessing, cleaning and feature creation

  - → [`src/data.py`](src/data.py)

- [`notebooks/02_encoder_decoder.ipynb`](notebooks/02_encoder_decoder.ipynb)

  - Prototype the encoder-decoder pretraining part of the model

  - → [`models/encoder_decoder.py`](models/encoder_decoder.py)

  - → [`src/utils.py`](src/utils.py)

- [`notebooks/03_encoder_decoder_dropout.ipynb`](notebooks/03_encoder_decoder_dropout.ipynb)

  - Prototype the encoder-decoder pretraining part of the model using [LSTM cells with dropout](https://github.com/keitakurita/Better_LSTM_PyTorch)

  - → [`models/encoder_decoder_dropout.py`](models/encoder_decoder_dropout.py)

- [`notebooks/04_pretraining_hyperparam.ipynb`](notebooks/04_pretraining_hyperparam.ipynb)

  - Use [`Ax`](https://github.com/facebook/Ax) for guided hyperparameter search in the pretraining of encoder-decoder. We use GCE compute for GPU acceleration

  - → [`src/utils.py`](src/utils.py)

- [`notebooks/05_pretraining_embedding.ipynb`](notebooks/05_pretraining_embedding.ipynb)

  - Taking a look at the learned pretraining embedding

  - → [`src/utils.py`](src/utils.py)

- [`notebooks/06_prediction_network.ipynb`](notebooks/06_prediction_network.ipynb)

  - Prototype the prediction network and run `Ax` for hyperparameter search

  - → [`src/utils.py`](src/utils.py)

  - → [`src/data.py`](src/data.py)

  - → [`models/prediction.py`](models/prediction.py)

- [`notebooks/07_full_inference.ipynb`](notebooks/07_full_inference.ipynb)

  - Implement the full inference including uncertainty calculation algorithms, run the test data through inference and plot results

  - → [`src/inference.py`](src/inference.py)

- [`notebooks/08_evaluation.ipynb`](notebooks/08_evaluation.ipynb)

  - Evaluate predictions on the test set by comparing with predictions from [facebook/prophet](https://github.com/facebook/prophet)

  - → [`src/evaluation.py`](src/evaluation.py)

## Results

In [`notebooks/08_evaluation.ipynb`](notebooks/08_evaluation.ipynb) Prediction results on the test set are compared to those made with [facebook/prophet](https://github.com/facebook/prophet). 

Results on the classical time series prediction evaluation metrics are presented below:

| Metric                                   |  Uber    | Prophet |

| ---------------------------------------- | -------- | ------- |

| Mean absolute error                      | 280.47   | 680.98  |

| Root mean squared error                  | 490.92   | 955.85  |

| Mean absolute percentage error           | 0.13     | 0.41    |

| Symmetric mean absolute percentage error | 0.029    | 0.024   |

Time series predicitons using the full inference algorithm including uncertainty bounds:

![Time series predictions with test set data](docs/results.png)

## Requirements

```

numpy

pandas

torch

tqdm

matplotlib

ax-platform

fb-prophet

```