This is a python
package containing Various Forecasting Algorithms,
Forecasting Datasets and, Plotting, Preprocessing and Utility Tools.
This package provides (or would provide) various algorithms which work
on data containing multivariate target time series, univariate target
time series - both with or without exogenous time series. The algorithms
are provided in the model
subpackage of the time series package ts
.
Currently, this package provides the following Forecasting Algorithms,
from ts.model import DeepNN
from ts.model import RnnForecast
from ts.model import SimpleRnnForecast
from ts.model import GruForecast
from ts.model import LstmForecast
from ts.model import ExtremeTime
from ts.model import ExtremeTime2
from ts.model import GmmHmmForecast
from ts.model import GmmHmmLikelihoodSimilarity
This package provides Data Generators as well as Datasets. The data
subpackage of the time series package ts
provides two subpackages
named generate
and dataset
, the first one contains data generators
and the second one contains real world datasets.
Currently, we provide the following,
from ts.model.univariate.nonexo import ArmaGenerator
from ts.model.univariate.nonexo import StandardGenerator
from ts.model.univariate.nonexo import PeriodicGenerator
from ts.model.univariate.nonexo import PolynomialGenerator
from ts.model.univariate.nonexo import DifficultGenerator
This package contains plotting tools for plotting losses, plotting training
data and comparing prediction with true (using plots). The plotting tools
are available in the plot
subpackage of the time series package ts
.
It also provides utility tools for in the utility
subpackage of ts
.
Access to the global logger and local loggers is provided in the log
subpackage of ts
.
ts/
|__ data/
| |__ dataset/
| |__ AmazonStockPrice
| |__ RetailSales
| |__ JaipurWeather
|
| |__ generate/
| |__ univariate/
| |__ nonexo/
| |__ ArmaGenerator
| |__ StandardGenerator
| |__ PeriodicGenerator
| |__ PolynomialGenerator
| |__ DifficultGenerator
|
|__ model/
| |__ RnnForecast
| |__ SimpleRnnForecast
| |__ GruForecast
| |__ LstmForecast
| |__ DeepNN
| |__ ExtremeTime
| |__ ExtremeTime2
| |__ GmmHmmForecast
| |__ GmmHmmLikelihoodSimilarity
|
|__ plot/
| |__ Plot
|
|__ log/
| |__ ConsoleLogger
| |__ FileLogger
| |__ GlobalLogger
|
|__ test/
| |__ model/..
| |__ utility/..
|
|__ utility/
| |__ Utility
| |__ ForecastDataSequence
| |__ SaveCallback
| |__ DatasetUtility
| |__ Metric
This repository is structured as follows:
Forecast
|__ other/...
|
|__ notebooks/..
|
|__ ts/...
other
: contains deprecated codes and codes that do not work.
The contents of this directory would be removed as soon as
they have been analyzed thoroughly.
notebooks
: contains notebooks containing experiments, tests and examples
ts
: The time series forecasting package
The notebooks
directory of this repository contains notebooks
containing experiments, tests and examples. To be able to run
these notebooks, we need to do as follows:
export PYTHONPATH="$PYTHONPATH:<location_of_this_repository>"
where <location_of_this_repository>
is the location of this repository
in your filesystem.