# LazyProphet v0.3.8
## Recent Changes
With v0.3.8 comes a fully fledged Optuna Optimizer for simple (no exogenous) regression problems. Classification is ToDo.
A Quick example of the new functionality:
```
from LazyProphet import LazyProphet as lp
from sklearn.datasets import fetch_openml
import matplotlib.pyplot as plt
bike_sharing = fetch_openml("Bike_Sharing_Demand", version=2, as_frame=True)
y = bike_sharing.frame['count']
y = y[-400:].values
lp_model = lp.LazyProphet.Optimize(y,
seasonal_period=[24, 168],
n_folds=2, # must be greater than 1
n_trials=20, # number of optimization runs, default is 100
test_size=48 # size of the holdout set to test against
)
fitted = lp_model.fit(y)
predicted = lp_model.predict(100)
plt.plot(y)
plt.plot(np.append(fitted, predicted))
plt.axvline(400)
plt.show()
```
## Introduction
[A decent intro can be found here.](https://medium.com/p/3745bafe5ce5)
LazyProphet is a time series forecasting model built for LightGBM forecasting of single time series.
Many nice-ities have been added such as recursive forecasting when using lagged target variable such as the last 4 values to predict the 5th.
Additionally, fourier basis functions and penalized weighted piecewise linear basis functions are options as well!
Don't ever use in-sample fit for these types of models as they fit the data quite snuggly.
## Quickstart
```
pip install LazyProphet
```
Simple example from Sklearn, just give it the hyperparameters and an array:
```
from LazyProphet import LazyProphet as lp
from sklearn.datasets import fetch_openml
import matplotlib.pyplot as plt
bike_sharing = fetch_openml("Bike_Sharing_Demand", version=2, as_frame=True)
y = bike_sharing.frame['count']
y = y[-400:].values
lp_model = lp.LazyProphet(seasonal_period=[24, 168], #list means we use both seasonal periods
n_basis=4, #weighted piecewise basis functions
fourier_order=10,
ar=list(range(1,25)),
decay=.99 #the 'penalized' in penalized weighted piecewise linear basis functions
)
fitted = lp_model.fit(y)
predicted = lp_model.predict(100)
plt.plot(y)
plt.plot(np.append(fitted, predicted))
plt.axvline(400)
plt.show()
```
If you are working with less data or then you will probably want to pass custom LightGBM params via boosting_params when creating the LazyProphet obj.
The default params are:
```
boosting_params = {
"objective": "regression",
"metric": "rmse",
"verbosity": -1,
"boosting_type": "gbdt",
"seed": 42,
'linear_tree': False,
'learning_rate': .15,
'min_child_samples': 5,
'num_leaves': 31,
'num_iterations': 50
}
```
*WARNING*
Passing linear_tree=True can be extremely unstable, especially with ar and n_basis arguments. We do tests for linearity and will de-trend if necessary.
**
Most importantly for controlling the complexity by using num_leaves/learning_rate for complexity with less data.
Alternatively, you could try out the method:
```
tree_optimize(y, exogenous=None, cv_splits=3, test_size=None)
```
In-place of the fit method. This will do 'cv_splits' number of Time-Series Cross-Validation steps to optimize the tree using Optuna. This method has some degraded performance in testing but may be better for autoforecasting various types of data sizes.
Raw data
{
"_id": null,
"home_page": "https://github.com/tblume1992/LazyProphet",
"name": "LazyProphet",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "forecasting,time series,lightgbm",
"author": "Tyler Blume",
"author_email": "tblume@mail.USF.edu",
"download_url": "",
"platform": null,
"description": "# LazyProphet v0.3.8\n\n## Recent Changes\n\nWith v0.3.8 comes a fully fledged Optuna Optimizer for simple (no exogenous) regression problems. Classification is ToDo.\n\nA Quick example of the new functionality:\n\n```\nfrom LazyProphet import LazyProphet as lp\nfrom sklearn.datasets import fetch_openml\nimport matplotlib.pyplot as plt\n\nbike_sharing = fetch_openml(\"Bike_Sharing_Demand\", version=2, as_frame=True)\ny = bike_sharing.frame['count']\ny = y[-400:].values\n\nlp_model = lp.LazyProphet.Optimize(y,\n seasonal_period=[24, 168],\n n_folds=2, # must be greater than 1\n n_trials=20, # number of optimization runs, default is 100\n test_size=48 # size of the holdout set to test against\n )\nfitted = lp_model.fit(y)\npredicted = lp_model.predict(100)\n\nplt.plot(y)\nplt.plot(np.append(fitted, predicted))\nplt.axvline(400)\nplt.show()\n```\n\n## Introduction\n\n[A decent intro can be found here.](https://medium.com/p/3745bafe5ce5)\n\nLazyProphet is a time series forecasting model built for LightGBM forecasting of single time series.\n\nMany nice-ities have been added such as recursive forecasting when using lagged target variable such as the last 4 values to predict the 5th.\n\nAdditionally, fourier basis functions and penalized weighted piecewise linear basis functions are options as well!\n\nDon't ever use in-sample fit for these types of models as they fit the data quite snuggly.\n\n## Quickstart\n\n```\npip install LazyProphet\n```\n\nSimple example from Sklearn, just give it the hyperparameters and an array:\n\n```\nfrom LazyProphet import LazyProphet as lp\nfrom sklearn.datasets import fetch_openml\nimport matplotlib.pyplot as plt\n\nbike_sharing = fetch_openml(\"Bike_Sharing_Demand\", version=2, as_frame=True)\ny = bike_sharing.frame['count']\ny = y[-400:].values\n\nlp_model = lp.LazyProphet(seasonal_period=[24, 168], #list means we use both seasonal periods\n n_basis=4, #weighted piecewise basis functions\n fourier_order=10,\n ar=list(range(1,25)),\n decay=.99 #the 'penalized' in penalized weighted piecewise linear basis functions\n )\nfitted = lp_model.fit(y)\npredicted = lp_model.predict(100)\n\nplt.plot(y)\nplt.plot(np.append(fitted, predicted))\nplt.axvline(400)\nplt.show()\n```\n![alt text](https://github.com/tblume1992/LazyProphet/blob/main/LazyProphet/static/example_output.png \"Output 1\")\n\nIf you are working with less data or then you will probably want to pass custom LightGBM params via boosting_params when creating the LazyProphet obj.\n\nThe default params are:\n\n```\nboosting_params = {\n \"objective\": \"regression\",\n \"metric\": \"rmse\",\n \"verbosity\": -1,\n \"boosting_type\": \"gbdt\",\n \"seed\": 42,\n 'linear_tree': False,\n 'learning_rate': .15,\n 'min_child_samples': 5,\n 'num_leaves': 31,\n 'num_iterations': 50\n }\n```\n*WARNING* \nPassing linear_tree=True can be extremely unstable, especially with ar and n_basis arguments. We do tests for linearity and will de-trend if necessary.\n**\n\nMost importantly for controlling the complexity by using num_leaves/learning_rate for complexity with less data.\n\nAlternatively, you could try out the method:\n\n```\ntree_optimize(y, exogenous=None, cv_splits=3, test_size=None)\n```\nIn-place of the fit method. This will do 'cv_splits' number of Time-Series Cross-Validation steps to optimize the tree using Optuna. This method has some degraded performance in testing but may be better for autoforecasting various types of data sizes.\n",
"bugtrack_url": null,
"license": "",
"summary": "Time series forecasting with LightGBM",
"version": "0.3.9",
"project_urls": {
"Homepage": "https://github.com/tblume1992/LazyProphet"
},
"split_keywords": [
"forecasting",
"time series",
"lightgbm"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "59d1935fdd0be9c2788857af1db3e7f2dad2fe967d39b0b75e5362899c8355a0",
"md5": "3c73cc44b3f838e64c7d8d2075a434e8",
"sha256": "1e297deb3c7ce035594c0644c65aea8daf908957c8f413784344687587f94e2e"
},
"downloads": -1,
"filename": "LazyProphet-0.3.9-py3-none-any.whl",
"has_sig": false,
"md5_digest": "3c73cc44b3f838e64c7d8d2075a434e8",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 10190,
"upload_time": "2023-08-14T16:11:12",
"upload_time_iso_8601": "2023-08-14T16:11:12.390816Z",
"url": "https://files.pythonhosted.org/packages/59/d1/935fdd0be9c2788857af1db3e7f2dad2fe967d39b0b75e5362899c8355a0/LazyProphet-0.3.9-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-08-14 16:11:12",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "tblume1992",
"github_project": "LazyProphet",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "lazyprophet"
}
JSON
