# tsai
<!-- WARNING: THIS FILE WAS AUTOGENERATED! DO NOT EDIT! -->
<div align="center">
<img src="https://github.com/timeseriesAI/tsai/blob/main/nbs/multimedia/tsai_logo.svg?raw=true" width="50%">
</div>
<br /> <br />
[](https://pypi.org/project/tsai/#description)
[](https://anaconda.org/timeseriesai/tsai)
[](https://zenodo.org/badge/latestdoi/211822289)
## Description
> State-of-the-art Deep Learning library for Time Series and Sequences.
`tsai` is an open-source deep learning package built on top of Pytorch &
fastai focused on state-of-the-art techniques for time series tasks like
classification, regression, forecasting, imputation…
`tsai` is currently under active development by timeseriesAI.
## What’s new:
During the last few releases, here are some of the most significant
additions to `tsai`:
- **New models**: PatchTST (Accepted by ICLR 2023), RNN with Attention
(RNNAttention, LSTMAttention, GRUAttention), TabFusionTransformer, …
- **New datasets**: we have increased the number of datasets you can
download using `tsai`:
- 128 univariate classification datasets
- 30 multivariate classification datasets
- 15 regression datasets
- 62 forecasting datasets
- 9 long term forecasting datasets
- **New tutorials**:
[PatchTST](https://github.com/timeseriesAI/tsai/blob/main/tutorial_nbs/15_PatchTST_a_new_transformer_for_LTSF.ipynb).
Based on some of your requests, we are planning to release additional
tutorials on data preparation and forecasting.
- **New functionality**: sklearn-type pipeline transforms, walk-foward
cross validation, reduced RAM requirements, and a lot of new
functionality to perform more accurate time series forecasts.
- Pytorch 2.0 support.
## Installation
### Pip install
You can install the **latest stable** version from pip using:
``` python
pip install tsai
```
If you plan to develop tsai yourself, or want to be on the cutting edge,
you can use an editable install. First install PyTorch, and then:
``` python
git clone https://github.com/timeseriesAI/tsai
pip install -e "tsai[dev]"
```
Note: starting with tsai 0.3.0 tsai will only install hard dependencies.
Other soft dependencies (which are only required for selected tasks)
will not be installed by default (this is the recommended approach. If
you require any of the dependencies that is not installed, tsai will ask
you to install it when necessary). If you still want to install tsai
with all its dependencies you can do it by running:
``` python
pip install tsai[extras]
```
### Conda install
You can also install tsai using conda (note that if you replace conda
with mamba the install process will be much faster and more reliable):
``` python
conda install -c timeseriesai tsai
```
## Documentation
Here’s the link to the
[documentation](https://timeseriesai.github.io/tsai/).
## Available models:
Here’s a list with some of the state-of-the-art models available in
`tsai`:
- [LSTM](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN.py)
(Hochreiter, 1997)
([paper](https://ieeexplore.ieee.org/abstract/document/6795963/))
- [GRU](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN.py)
(Cho, 2014) ([paper](https://arxiv.org/abs/1412.3555))
- [MLP](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/MLP.py) -
Multilayer Perceptron (Wang, 2016)
([paper](https://arxiv.org/abs/1611.06455))
- [FCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/FCN.py) -
Fully Convolutional Network (Wang, 2016)
([paper](https://arxiv.org/abs/1611.06455))
- [ResNet](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/ResNet.py) -
Residual Network (Wang, 2016)
([paper](https://arxiv.org/abs/1611.06455))
- [LSTM-FCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN_FCN.py)
(Karim, 2017) ([paper](https://arxiv.org/abs/1709.05206))
- [GRU-FCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN_FCN.py)
(Elsayed, 2018) ([paper](https://arxiv.org/abs/1812.07683))
- [mWDN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/mWDN.py) -
Multilevel wavelet decomposition network (Wang, 2018)
([paper](https://dl.acm.org/doi/abs/10.1145/3219819.3220060))
- [TCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TCN.py) -
Temporal Convolutional Network (Bai, 2018)
([paper](https://arxiv.org/abs/1803.01271))
- [MLSTM-FCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN_FCN.py) -
Multivariate LSTM-FCN (Karim, 2019)
([paper](https://www.sciencedirect.com/science/article/abs/pii/S0893608019301200))
- [InceptionTime](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/InceptionTime.py)
(Fawaz, 2019) ([paper](https://arxiv.org/abs/1909.04939))
- [Rocket](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/ROCKET.py)
(Dempster, 2019) ([paper](https://arxiv.org/abs/1910.13051))
- [XceptionTime](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/XceptionTime.py)
(Rahimian, 2019) ([paper](https://arxiv.org/abs/1911.03803))
- [ResCNN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/ResCNN.py) -
1D-ResCNN (Zou , 2019)
([paper](https://www.sciencedirect.com/science/article/pii/S0925231219311506))
- [TabModel](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TabModel.py) -
modified from fastai’s
[TabularModel](https://docs.fast.ai/tabular.model.html#TabularModel)
- [OmniScale](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/OmniScaleCNN.py) -
Omni-Scale 1D-CNN (Tang, 2020)
([paper](https://arxiv.org/abs/2002.10061))
- [TST](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TST.py) -
Time Series Transformer (Zerveas, 2020)
([paper](https://dl.acm.org/doi/abs/10.1145/3447548.3467401))
- [TabTransformer](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TabTransformer.py)
(Huang, 2020) ([paper](https://arxiv.org/pdf/2012.06678))
- [TSiT](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TSiTPlus.py)
Adapted from ViT (Dosovitskiy, 2020)
([paper](https://arxiv.org/abs/2010.11929))
- [MiniRocket](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/MINIROCKET.py)
(Dempster, 2021) ([paper](https://arxiv.org/abs/2102.00457))
- [XCM](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/XCM.py) -
An Explainable Convolutional Neural Network (Fauvel, 2021)
([paper](https://hal.inria.fr/hal-03469487/document))
- [gMLP](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/gMLP.py) -
Gated Multilayer Perceptron (Liu, 2021)
([paper](https://arxiv.org/abs/2105.08050))
- [TSPerceiver](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TSPerceiver.py) -
Adapted from Perceiver IO (Jaegle, 2021)
([paper](https://arxiv.org/abs/2107.14795))
- [GatedTabTransformer](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/GatedTabTransformer.py)
(Cholakov, 2022) ([paper](https://arxiv.org/abs/2201.00199))
- [TSSequencerPlus](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TSSequencerPlus.py) -
Adapted from Sequencer (Tatsunami, 2022)
([paper](https://arxiv.org/abs/2205.01972))
- [PatchTST](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/PatchTST.py) -
(Nie, 2022) ([paper](https://arxiv.org/abs/2211.14730))
plus other custom models like: TransformerModel, LSTMAttention,
GRUAttention, …
## How to start using tsai?
To get to know the tsai package, we’d suggest you start with this
notebook in Google Colab:
**[01_Intro_to_Time_Series_Classification](https://colab.research.google.com/github/timeseriesAI/tsai/blob/master/tutorial_nbs/01_Intro_to_Time_Series_Classification.ipynb)**
It provides an overview of a time series classification task.
We have also develop many other [tutorial
notebooks](https://github.com/timeseriesAI/tsai/tree/main/tutorial_nbs).
To use tsai in your own notebooks, the only thing you need to do after
you have installed the package is to run this:
``` python
from tsai.all import *
```
## Examples
These are just a few examples of how you can use `tsai`:
### Binary, univariate classification
**Training:**
``` python
from tsai.basics import *
X, y, splits = get_classification_data('ECG200', split_data=False)
tfms = [None, TSClassification()]
batch_tfms = TSStandardize()
clf = TSClassifier(X, y, splits=splits, path='models', arch="InceptionTimePlus", tfms=tfms, batch_tfms=batch_tfms, metrics=accuracy, cbs=ShowGraph())
clf.fit_one_cycle(100, 3e-4)
clf.export("clf.pkl")
```
**Inference:**
``` python
from tsai.inference import load_learner
clf = load_learner("models/clf.pkl")
probas, target, preds = clf.get_X_preds(X[splits[1]], y[splits[1]])
```
### Multi-class, multivariate classification
**Training:**
``` python
from tsai.basics import *
X, y, splits = get_classification_data('LSST', split_data=False)
tfms = [None, TSClassification()]
batch_tfms = TSStandardize(by_sample=True)
mv_clf = TSClassifier(X, y, splits=splits, path='models', arch="InceptionTimePlus", tfms=tfms, batch_tfms=batch_tfms, metrics=accuracy, cbs=ShowGraph())
mv_clf.fit_one_cycle(10, 1e-2)
mv_clf.export("mv_clf.pkl")
```
**Inference:**
``` python
from tsai.inference import load_learner
mv_clf = load_learner("models/mv_clf.pkl")
probas, target, preds = mv_clf.get_X_preds(X[splits[1]], y[splits[1]])
```
### Multivariate Regression
**Training:**
``` python
from tsai.basics import *
X, y, splits = get_regression_data('AppliancesEnergy', split_data=False)
tfms = [None, TSRegression()]
batch_tfms = TSStandardize(by_sample=True)
reg = TSRegressor(X, y, splits=splits, path='models', arch="TSTPlus", tfms=tfms, batch_tfms=batch_tfms, metrics=rmse, cbs=ShowGraph(), verbose=True)
reg.fit_one_cycle(100, 3e-4)
reg.export("reg.pkl")
```
**Inference:**
``` python
from tsai.inference import load_learner
reg = load_learner("models/reg.pkl")
raw_preds, target, preds = reg.get_X_preds(X[splits[1]], y[splits[1]])
```
The ROCKETs (RocketClassifier, RocketRegressor, MiniRocketClassifier,
MiniRocketRegressor, MiniRocketVotingClassifier or
MiniRocketVotingRegressor) are somewhat different models. They are not
actually deep learning models (although they use convolutions) and are
used in a different way.
⚠️ You’ll also need to install sktime to be able to use them. You can
install it separately:
``` python
pip install sktime
```
or use:
``` python
pip install tsai[extras]
```
**Training:**
``` python
from sklearn.metrics import mean_squared_error, make_scorer
from tsai.data.external import get_Monash_regression_data
from tsai.models.MINIROCKET import MiniRocketRegressor
X_train, y_train, *_ = get_Monash_regression_data('AppliancesEnergy')
rmse_scorer = make_scorer(mean_squared_error, greater_is_better=False)
reg = MiniRocketRegressor(scoring=rmse_scorer)
reg.fit(X_train, y_train)
reg.save('MiniRocketRegressor')
```
**Inference:**
``` python
from sklearn.metrics import mean_squared_error
from tsai.data.external import get_Monash_regression_data
from tsai.models.MINIROCKET import load_minirocket
*_, X_test, y_test = get_Monash_regression_data('AppliancesEnergy')
reg = load_minirocket('MiniRocketRegressor')
y_pred = reg.predict(X_test)
mean_squared_error(y_test, y_pred, squared=False)
```
### Forecasting
You can use tsai for forecast in the following scenarios:
- univariate or multivariate time series input
- univariate or multivariate time series output
- single or multi-step ahead
You’ll need to: \* prepare X (time series input) and the target y (see
[documentation](https://timeseriesai.github.io/tsai/data.preparation.html))
\* select PatchTST or one of tsai’s models ending in Plus (TSTPlus,
InceptionTimePlus, TSiTPlus, etc). The model will auto-configure a head
to yield an output with the same shape as the target input y.
#### Single step
**Training:**
``` python
from tsai.basics import *
ts = get_forecasting_time_series("Sunspots").values
X, y = SlidingWindow(60, horizon=1)(ts)
splits = TimeSplitter(235)(y)
tfms = [None, TSForecasting()]
batch_tfms = TSStandardize()
fcst = TSForecaster(X, y, splits=splits, path='models', tfms=tfms, batch_tfms=batch_tfms, bs=512, arch="TSTPlus", metrics=mae, cbs=ShowGraph())
fcst.fit_one_cycle(50, 1e-3)
fcst.export("fcst.pkl")
```
**Inference:**
``` python
from tsai.inference import load_learner
fcst = load_learner("models/fcst.pkl", cpu=False)
raw_preds, target, preds = fcst.get_X_preds(X[splits[1]], y[splits[1]])
raw_preds.shape
# torch.Size([235, 1])
```
#### Multi-step
This example show how to build a 3-step ahead univariate forecast.
**Training:**
``` python
from tsai.basics import *
ts = get_forecasting_time_series("Sunspots").values
X, y = SlidingWindow(60, horizon=3)(ts)
splits = TimeSplitter(235, fcst_horizon=3)(y)
tfms = [None, TSForecasting()]
batch_tfms = TSStandardize()
fcst = TSForecaster(X, y, splits=splits, path='models', tfms=tfms, batch_tfms=batch_tfms, bs=512, arch="TSTPlus", metrics=mae, cbs=ShowGraph())
fcst.fit_one_cycle(50, 1e-3)
fcst.export("fcst.pkl")
```
**Inference:**
``` python
from tsai.inference import load_learner
fcst = load_learner("models/fcst.pkl", cpu=False)
raw_preds, target, preds = fcst.get_X_preds(X[splits[1]], y[splits[1]])
raw_preds.shape
# torch.Size([235, 3])
```
## Input data format
The input format for all time series models and image models in tsai is
the same. An np.ndarray (or array-like object like zarr, etc) with 3
dimensions:
**\[# samples x \# variables x sequence length\]**
The input format for tabular models in tsai (like TabModel,
TabTransformer and TabFusionTransformer) is a pandas dataframe. See
[example](https://timeseriesai.github.io/tsai/models.TabModel.html).
## How to contribute to tsai?
We welcome contributions of all kinds. Development of enhancements, bug
fixes, documentation, tutorial notebooks, …
We have created a guide to help you start contributing to tsai. You can
read it
[here](https://github.com/timeseriesAI/tsai/blob/main/CONTRIBUTING.md).
## Enterprise support and consulting services:
Want to make the most out of timeseriesAI/tsai in a professional
setting? Let us help. Send us an email to learn more:
info@timeseriesai.co
## Citing tsai
If you use tsai in your research please use the following BibTeX entry:
``` text
@Misc{tsai,
author = {Ignacio Oguiza},
title = {tsai - A state-of-the-art deep learning library for time series and sequential data},
howpublished = {Github},
year = {2023},
url = {https://github.com/timeseriesAI/tsai}
}
```
Raw data
{
"_id": null,
"home_page": "https://github.com/timeseriesAI/tsai/",
"name": "tsai",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": "",
"keywords": "fastai time-series time-series-classification time-series-regression deep-learning Pytorch",
"author": "Ignacio Oguiza and contributors",
"author_email": "oguiza@timeseriesAI.co",
"download_url": "https://files.pythonhosted.org/packages/24/aa/c0ef2a57488a7d635356d7bb964cf57350f28024d723ac5721007d4a596a/tsai-0.3.9.tar.gz",
"platform": null,
"description": "# tsai\n\n<!-- WARNING: THIS FILE WAS AUTOGENERATED! DO NOT EDIT! -->\n\n<div align=\"center\">\n\n<img src=\"https://github.com/timeseriesAI/tsai/blob/main/nbs/multimedia/tsai_logo.svg?raw=true\" width=\"50%\">\n\n</div>\n\n<br /> <br />\n\n\n[](https://pypi.org/project/tsai/#description)\n[](https://anaconda.org/timeseriesai/tsai)\n[](https://zenodo.org/badge/latestdoi/211822289)\n\n\n## Description\n\n> State-of-the-art Deep Learning library for Time Series and Sequences.\n\n`tsai` is an open-source deep learning package built on top of Pytorch &\nfastai focused on state-of-the-art techniques for time series tasks like\nclassification, regression, forecasting, imputation\u2026\n\n`tsai` is currently under active development by timeseriesAI.\n\n## What\u2019s new:\n\nDuring the last few releases, here are some of the most significant\nadditions to `tsai`:\n\n- **New models**: PatchTST (Accepted by ICLR 2023), RNN with Attention\n (RNNAttention, LSTMAttention, GRUAttention), TabFusionTransformer, \u2026\n- **New datasets**: we have increased the number of datasets you can\n download using `tsai`:\n - 128 univariate classification datasets\n - 30 multivariate classification datasets\n - 15 regression datasets\n - 62 forecasting datasets\n - 9 long term forecasting datasets\n- **New tutorials**:\n [PatchTST](https://github.com/timeseriesAI/tsai/blob/main/tutorial_nbs/15_PatchTST_a_new_transformer_for_LTSF.ipynb).\n Based on some of your requests, we are planning to release additional\n tutorials on data preparation and forecasting.\n- **New functionality**: sklearn-type pipeline transforms, walk-foward\n cross validation, reduced RAM requirements, and a lot of new\n functionality to perform more accurate time series forecasts.\n- Pytorch 2.0 support.\n\n## Installation\n\n### Pip install\n\nYou can install the **latest stable** version from pip using:\n\n``` python\npip install tsai\n```\n\nIf you plan to develop tsai yourself, or want to be on the cutting edge,\nyou can use an editable install. First install PyTorch, and then:\n\n``` python\ngit clone https://github.com/timeseriesAI/tsai\npip install -e \"tsai[dev]\"\n```\n\nNote: starting with tsai 0.3.0 tsai will only install hard dependencies.\nOther soft dependencies (which are only required for selected tasks)\nwill not be installed by default (this is the recommended approach. If\nyou require any of the dependencies that is not installed, tsai will ask\nyou to install it when necessary). If you still want to install tsai\nwith all its dependencies you can do it by running:\n\n``` python\npip install tsai[extras]\n```\n\n### Conda install\n\nYou can also install tsai using conda (note that if you replace conda\nwith mamba the install process will be much faster and more reliable):\n\n``` python\nconda install -c timeseriesai tsai\n```\n\n## Documentation\n\nHere\u2019s the link to the\n[documentation](https://timeseriesai.github.io/tsai/).\n\n## Available models:\n\nHere\u2019s a list with some of the state-of-the-art models available in\n`tsai`:\n\n- [LSTM](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN.py)\n (Hochreiter, 1997)\n ([paper](https://ieeexplore.ieee.org/abstract/document/6795963/))\n- [GRU](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN.py)\n (Cho, 2014) ([paper](https://arxiv.org/abs/1412.3555))\n- [MLP](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/MLP.py) -\n Multilayer Perceptron (Wang, 2016)\n ([paper](https://arxiv.org/abs/1611.06455))\n- [FCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/FCN.py) -\n Fully Convolutional Network (Wang, 2016)\n ([paper](https://arxiv.org/abs/1611.06455))\n- [ResNet](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/ResNet.py) -\n Residual Network (Wang, 2016)\n ([paper](https://arxiv.org/abs/1611.06455))\n- [LSTM-FCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN_FCN.py)\n (Karim, 2017) ([paper](https://arxiv.org/abs/1709.05206))\n- [GRU-FCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN_FCN.py)\n (Elsayed, 2018) ([paper](https://arxiv.org/abs/1812.07683))\n- [mWDN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/mWDN.py) -\n Multilevel wavelet decomposition network (Wang, 2018)\n ([paper](https://dl.acm.org/doi/abs/10.1145/3219819.3220060))\n- [TCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TCN.py) -\n Temporal Convolutional Network (Bai, 2018)\n ([paper](https://arxiv.org/abs/1803.01271))\n- [MLSTM-FCN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/RNN_FCN.py) -\n Multivariate LSTM-FCN (Karim, 2019)\n ([paper](https://www.sciencedirect.com/science/article/abs/pii/S0893608019301200))\n- [InceptionTime](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/InceptionTime.py)\n (Fawaz, 2019) ([paper](https://arxiv.org/abs/1909.04939))\n- [Rocket](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/ROCKET.py)\n (Dempster, 2019) ([paper](https://arxiv.org/abs/1910.13051))\n- [XceptionTime](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/XceptionTime.py)\n (Rahimian, 2019) ([paper](https://arxiv.org/abs/1911.03803))\n- [ResCNN](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/ResCNN.py) -\n 1D-ResCNN (Zou , 2019)\n ([paper](https://www.sciencedirect.com/science/article/pii/S0925231219311506))\n- [TabModel](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TabModel.py) -\n modified from fastai\u2019s\n [TabularModel](https://docs.fast.ai/tabular.model.html#TabularModel)\n- [OmniScale](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/OmniScaleCNN.py) -\n Omni-Scale 1D-CNN (Tang, 2020)\n ([paper](https://arxiv.org/abs/2002.10061))\n- [TST](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TST.py) -\n Time Series Transformer (Zerveas, 2020)\n ([paper](https://dl.acm.org/doi/abs/10.1145/3447548.3467401))\n- [TabTransformer](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TabTransformer.py)\n (Huang, 2020) ([paper](https://arxiv.org/pdf/2012.06678))\n- [TSiT](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TSiTPlus.py)\n Adapted from ViT (Dosovitskiy, 2020)\n ([paper](https://arxiv.org/abs/2010.11929))\n- [MiniRocket](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/MINIROCKET.py)\n (Dempster, 2021) ([paper](https://arxiv.org/abs/2102.00457))\n- [XCM](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/XCM.py) -\n An Explainable Convolutional Neural Network (Fauvel, 2021)\n ([paper](https://hal.inria.fr/hal-03469487/document))\n- [gMLP](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/gMLP.py) -\n Gated Multilayer Perceptron (Liu, 2021)\n ([paper](https://arxiv.org/abs/2105.08050))\n- [TSPerceiver](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TSPerceiver.py) -\n Adapted from Perceiver IO (Jaegle, 2021)\n ([paper](https://arxiv.org/abs/2107.14795))\n- [GatedTabTransformer](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/GatedTabTransformer.py)\n (Cholakov, 2022) ([paper](https://arxiv.org/abs/2201.00199))\n- [TSSequencerPlus](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/TSSequencerPlus.py) -\n Adapted from Sequencer (Tatsunami, 2022)\n ([paper](https://arxiv.org/abs/2205.01972))\n- [PatchTST](https://github.com/timeseriesAI/tsai/blob/main/tsai/models/PatchTST.py) -\n (Nie, 2022) ([paper](https://arxiv.org/abs/2211.14730))\n\nplus other custom models like: TransformerModel, LSTMAttention,\nGRUAttention, \u2026\n\n## How to start using tsai?\n\nTo get to know the tsai package, we\u2019d suggest you start with this\nnotebook in Google Colab:\n**[01_Intro_to_Time_Series_Classification](https://colab.research.google.com/github/timeseriesAI/tsai/blob/master/tutorial_nbs/01_Intro_to_Time_Series_Classification.ipynb)**\nIt provides an overview of a time series classification task.\n\nWe have also develop many other [tutorial\nnotebooks](https://github.com/timeseriesAI/tsai/tree/main/tutorial_nbs).\n\nTo use tsai in your own notebooks, the only thing you need to do after\nyou have installed the package is to run this:\n\n``` python\nfrom tsai.all import *\n```\n\n## Examples\n\nThese are just a few examples of how you can use `tsai`:\n\n### Binary, univariate classification\n\n**Training:**\n\n``` python\nfrom tsai.basics import *\n\nX, y, splits = get_classification_data('ECG200', split_data=False)\ntfms = [None, TSClassification()]\nbatch_tfms = TSStandardize()\nclf = TSClassifier(X, y, splits=splits, path='models', arch=\"InceptionTimePlus\", tfms=tfms, batch_tfms=batch_tfms, metrics=accuracy, cbs=ShowGraph())\nclf.fit_one_cycle(100, 3e-4)\nclf.export(\"clf.pkl\") \n```\n\n**Inference:**\n\n``` python\nfrom tsai.inference import load_learner\n\nclf = load_learner(\"models/clf.pkl\")\nprobas, target, preds = clf.get_X_preds(X[splits[1]], y[splits[1]])\n```\n\n### Multi-class, multivariate classification\n\n**Training:**\n\n``` python\nfrom tsai.basics import *\n\nX, y, splits = get_classification_data('LSST', split_data=False)\ntfms = [None, TSClassification()]\nbatch_tfms = TSStandardize(by_sample=True)\nmv_clf = TSClassifier(X, y, splits=splits, path='models', arch=\"InceptionTimePlus\", tfms=tfms, batch_tfms=batch_tfms, metrics=accuracy, cbs=ShowGraph())\nmv_clf.fit_one_cycle(10, 1e-2)\nmv_clf.export(\"mv_clf.pkl\")\n```\n\n**Inference:**\n\n``` python\nfrom tsai.inference import load_learner\n\nmv_clf = load_learner(\"models/mv_clf.pkl\")\nprobas, target, preds = mv_clf.get_X_preds(X[splits[1]], y[splits[1]])\n```\n\n### Multivariate Regression\n\n**Training:**\n\n``` python\nfrom tsai.basics import *\n\nX, y, splits = get_regression_data('AppliancesEnergy', split_data=False)\ntfms = [None, TSRegression()]\nbatch_tfms = TSStandardize(by_sample=True)\nreg = TSRegressor(X, y, splits=splits, path='models', arch=\"TSTPlus\", tfms=tfms, batch_tfms=batch_tfms, metrics=rmse, cbs=ShowGraph(), verbose=True)\nreg.fit_one_cycle(100, 3e-4)\nreg.export(\"reg.pkl\")\n```\n\n**Inference:**\n\n``` python\nfrom tsai.inference import load_learner\n\nreg = load_learner(\"models/reg.pkl\")\nraw_preds, target, preds = reg.get_X_preds(X[splits[1]], y[splits[1]])\n```\n\nThe ROCKETs (RocketClassifier, RocketRegressor, MiniRocketClassifier,\nMiniRocketRegressor, MiniRocketVotingClassifier or\nMiniRocketVotingRegressor) are somewhat different models. They are not\nactually deep learning models (although they use convolutions) and are\nused in a different way.\n\n\u26a0\ufe0f You\u2019ll also need to install sktime to be able to use them. You can\ninstall it separately:\n\n``` python\npip install sktime\n```\n\nor use:\n\n``` python\npip install tsai[extras]\n```\n\n**Training:**\n\n``` python\nfrom sklearn.metrics import mean_squared_error, make_scorer\nfrom tsai.data.external import get_Monash_regression_data\nfrom tsai.models.MINIROCKET import MiniRocketRegressor\n\nX_train, y_train, *_ = get_Monash_regression_data('AppliancesEnergy')\nrmse_scorer = make_scorer(mean_squared_error, greater_is_better=False)\nreg = MiniRocketRegressor(scoring=rmse_scorer)\nreg.fit(X_train, y_train)\nreg.save('MiniRocketRegressor')\n```\n\n**Inference:**\n\n``` python\nfrom sklearn.metrics import mean_squared_error\nfrom tsai.data.external import get_Monash_regression_data\nfrom tsai.models.MINIROCKET import load_minirocket\n\n*_, X_test, y_test = get_Monash_regression_data('AppliancesEnergy')\nreg = load_minirocket('MiniRocketRegressor')\ny_pred = reg.predict(X_test)\nmean_squared_error(y_test, y_pred, squared=False)\n```\n\n### Forecasting\n\nYou can use tsai for forecast in the following scenarios:\n\n- univariate or multivariate time series input\n- univariate or multivariate time series output\n- single or multi-step ahead\n\nYou\u2019ll need to: \\* prepare X (time series input) and the target y (see\n[documentation](https://timeseriesai.github.io/tsai/data.preparation.html))\n\\* select PatchTST or one of tsai\u2019s models ending in Plus (TSTPlus,\nInceptionTimePlus, TSiTPlus, etc). The model will auto-configure a head\nto yield an output with the same shape as the target input y.\n\n#### Single step\n\n**Training:**\n\n``` python\nfrom tsai.basics import *\n\nts = get_forecasting_time_series(\"Sunspots\").values\nX, y = SlidingWindow(60, horizon=1)(ts)\nsplits = TimeSplitter(235)(y) \ntfms = [None, TSForecasting()]\nbatch_tfms = TSStandardize()\nfcst = TSForecaster(X, y, splits=splits, path='models', tfms=tfms, batch_tfms=batch_tfms, bs=512, arch=\"TSTPlus\", metrics=mae, cbs=ShowGraph())\nfcst.fit_one_cycle(50, 1e-3)\nfcst.export(\"fcst.pkl\")\n```\n\n**Inference:**\n\n``` python\nfrom tsai.inference import load_learner\n\nfcst = load_learner(\"models/fcst.pkl\", cpu=False)\nraw_preds, target, preds = fcst.get_X_preds(X[splits[1]], y[splits[1]])\nraw_preds.shape\n# torch.Size([235, 1])\n```\n\n#### Multi-step\n\nThis example show how to build a 3-step ahead univariate forecast.\n\n**Training:**\n\n``` python\nfrom tsai.basics import *\n\nts = get_forecasting_time_series(\"Sunspots\").values\nX, y = SlidingWindow(60, horizon=3)(ts)\nsplits = TimeSplitter(235, fcst_horizon=3)(y) \ntfms = [None, TSForecasting()]\nbatch_tfms = TSStandardize()\nfcst = TSForecaster(X, y, splits=splits, path='models', tfms=tfms, batch_tfms=batch_tfms, bs=512, arch=\"TSTPlus\", metrics=mae, cbs=ShowGraph())\nfcst.fit_one_cycle(50, 1e-3)\nfcst.export(\"fcst.pkl\")\n```\n\n**Inference:**\n\n``` python\nfrom tsai.inference import load_learner\nfcst = load_learner(\"models/fcst.pkl\", cpu=False)\nraw_preds, target, preds = fcst.get_X_preds(X[splits[1]], y[splits[1]])\nraw_preds.shape\n# torch.Size([235, 3])\n```\n\n## Input data format\n\nThe input format for all time series models and image models in tsai is\nthe same. An np.ndarray (or array-like object like zarr, etc) with 3\ndimensions:\n\n**\\[# samples x \\# variables x sequence length\\]**\n\nThe input format for tabular models in tsai (like TabModel,\nTabTransformer and TabFusionTransformer) is a pandas dataframe. See\n[example](https://timeseriesai.github.io/tsai/models.TabModel.html).\n\n## How to contribute to tsai?\n\nWe welcome contributions of all kinds. Development of enhancements, bug\nfixes, documentation, tutorial notebooks, \u2026\n\nWe have created a guide to help you start contributing to tsai. You can\nread it\n[here](https://github.com/timeseriesAI/tsai/blob/main/CONTRIBUTING.md).\n\n## Enterprise support and consulting services:\n\nWant to make the most out of timeseriesAI/tsai in a professional\nsetting? Let us help. Send us an email to learn more:\ninfo@timeseriesai.co\n\n## Citing tsai\n\nIf you use tsai in your research please use the following BibTeX entry:\n\n``` text\n@Misc{tsai,\n author = {Ignacio Oguiza},\n title = {tsai - A state-of-the-art deep learning library for time series and sequential data},\n howpublished = {Github},\n year = {2023},\n url = {https://github.com/timeseriesAI/tsai}\n}\n```\n",
"bugtrack_url": null,
"license": "Apache Software License 2.0",
"summary": "Practical Deep Learning for Time Series / Sequential Data library based on fastai & Pytorch",
"version": "0.3.9",
"project_urls": {
"Documentation": "https://timeseriesAI.github.io/tsai/",
"Homepage": "https://github.com/timeseriesAI/tsai/"
},
"split_keywords": [
"fastai",
"time-series",
"time-series-classification",
"time-series-regression",
"deep-learning",
"pytorch"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "c4382e9a9989d623ad158d5b927cb157aa510e1eb8fe6903baf66985cb9ed950",
"md5": "e7d001b17cb04ab3dd74341fb2697f65",
"sha256": "0e6460e9ff075176c8af25ef90444ef2e287bc5fe5510e2a0f427e4d0b37ebc0"
},
"downloads": -1,
"filename": "tsai-0.3.9-py3-none-any.whl",
"has_sig": false,
"md5_digest": "e7d001b17cb04ab3dd74341fb2697f65",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 324340,
"upload_time": "2024-02-11T19:23:14",
"upload_time_iso_8601": "2024-02-11T19:23:14.959454Z",
"url": "https://files.pythonhosted.org/packages/c4/38/2e9a9989d623ad158d5b927cb157aa510e1eb8fe6903baf66985cb9ed950/tsai-0.3.9-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "24aac0ef2a57488a7d635356d7bb964cf57350f28024d723ac5721007d4a596a",
"md5": "992786e598873b32eaba5228c0107609",
"sha256": "514291b9ef3b7a8a80db704b9efcf9ae68ff43392a56fec674ecba46426004eb"
},
"downloads": -1,
"filename": "tsai-0.3.9.tar.gz",
"has_sig": false,
"md5_digest": "992786e598873b32eaba5228c0107609",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 280256,
"upload_time": "2024-02-11T19:23:17",
"upload_time_iso_8601": "2024-02-11T19:23:17.341707Z",
"url": "https://files.pythonhosted.org/packages/24/aa/c0ef2a57488a7d635356d7bb964cf57350f28024d723ac5721007d4a596a/tsai-0.3.9.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-02-11 19:23:17",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "timeseriesAI",
"github_project": "tsai",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "tsai"
}
JSON
