sketulate 1.0.0

Sketchable function and density simulations for data science

Sketulate

Sketulate is a Python package for interactive function, surface, and density simulation directly in Jupyter notebooks. Draw functions, surfaces, or densities with your mouse, convert them into callable functions or density distributions, and simulate data for experiments, modeling, or teaching.

Features

• Sketch univariate functions interactively (Sketulate)
• Sketch 2D surfaces / interactions effects (SketulateInteraction)
• Sketch density distributions
• Generates easy to use functions and sampleable density distributions
• Works seamlessly in Jupyter Notebook
• Easy integration for synthetic data generation

Installation

pip install sketulate

Quick Example

from sketulate import Sketulate, SketulateInteraction

Draw a univariate function

f1 = Sketulate() f1.sketch()

![Sketch a Function](examples/images/draw_a_function.png)

```python
# F1 is a now a ready to use function via:
f1.f
# f takes the form of a sklearn pipeline object,so you need to call predict on new data:
import numpy as np
x = np.linspace(0,1,100)
f1.f.predict(x.reshape(-1,1))
# Or a custom density distribution (selected in the canvas dropdown) via:
f1.g
# Draw 100 samples from g  by calling:
f1.g(100)

# Draw an interaction surface
f3 = SketulateInteraction(x_range=(0,10), y_range=(0,10), z_range=(-5,5), grid_size=5)
f3.sketch()  # Interactive surface with sliders

Put it all together and easily simulate some data

Saving and loading state

You can save a Sketulate (or SketulateInteraction) state to a dictionary and persist it as JSON, then restore it later.

1. Save the state from your sketulate object:

from sketulate import Sketulate
import json

sk = Sketulate()
sk.sketch()
# ... draw and accept a function or density ...

state = sk.save_state()

2. Write the state to a JSON file:

with open("sketulate_state.json", "w") as f:
    json.dump(state, f, indent=2)

3. Load the state from JSON (e.g. in a new session or notebook):

with open("sketulate_state.json", "r") as f:
    state = json.load(f)

4. Load the state into a Sketulate object and restore the fitted function or density:

sk_restored = Sketulate()
sk_restored.load_state(state)
# sk_restored.f (function) or sk_restored.g (density) is now available

Technical Note

For this version the sketches are modelled via piecewise linear basis functions using sklearn. Interaction surfaces are modelled using the LinearND interpolator from scipy. In addition, linear extrapolation is, by default, provided outside of the given ranges. Careful!