Datasets (reservoirpy.datasets
)#
Chaotic timeseries on continuous time#
Timeseries expressing a chaotic behaviour and defined on a continuous time axis, generated at will.
All timeseries defined by differential equations on a
continuous space are by default approximated using 4-5th order
Runge-Kuta method [1], either homemade (for Mackey-Glass timeseries)
or from Scipy scipy.integrate.solve_ivp()
tool.
|
Mackey-Glass timeseries [8] [9], computed from the Mackey-Glass delayed differential equation. |
|
Lorenz attractor timeseries as defined by Lorenz in 1963 [6] [7]. |
|
Double scroll attractor timeseries [10] [11], a particular case of multiscroll attractor timeseries. |
|
Double scroll attractor timeseries [10] [11], a particular case of multiscroll attractor timeseries. |
|
|
|
Lorenz96 attractor timeseries as defined by Lorenz in 1996 [17]. |
|
Rössler attractor timeseries [18]. |
|
Chaotic timeseries on discrete time#
Timeseries expressing a chaotic behaviour and defined on a discrete time axis, generated at will.
Discrete timeseries are defined using recurrent time-delay relations.
|
|
|
|
|
Non-linear Autoregressive Moving Average (NARMA) timeseries, as first defined in [14], and as used in [15]. |
Classification/pattern recognition tasks#
Classified datasets of timeseries.
|
Load the Japanese vowels [16] dataset. |
Miscellaneous#
|
Split a timeseries for forecasting tasks. |
|
Change the default random seed value used for dataset generation. |
|
Return the current random state seed used for dataset generation. |
References