as opposed to Random Number Generation
arange:
the difference between arange and linspace is that arange gets all values between start and stop with defined stepsize, while np.linspace gets a defined number of values. linspace infers the step size
{python} np.arange(start=0, stop=10, step=1)
it stops before the stop value.
example:
{python} np.arange(0, 10, 2) # [0 2 4 6 8]
linspace
As mentioned above, linspace infers the stepsize
{python} np.linspace(start=0, stop=10, num=20)
Note: set {python}endpoint=False if you don't want the endpoint.
Dataset generation
When working with neural networks, we want to test them using dummy data. This has the following goals:
- Make sure the model can handle the input data shape
Note that this can be achieved by generating truly random data Random Number Generation. - Make sure the model is able to detect basic data dependencies.
This is where we require more complex data generation functions, where the data isn't random anymore.
https://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_classification.html