Learn Machine Learning in Grasshopper - 3D Interactive Study Guide | 12-Step C# Script | Rhino 7/8
+======================================================================+
| PAYHIP PRODUCT LISTING |
| Learn Machine Learning in Grasshopper |
| 2026 Wickerson Studios |
+======================================================================+
------------------------------------------------------------------------
TITLE
------------------------------------------------------------------------
Learn Machine Learning in Grasshopper - 3D Interactive Study Guide |
12-Step C# Script | Rhino 7/8
------------------------------------------------------------------------
SUBTITLE / TAGLINE
------------------------------------------------------------------------
Data points scatter through 3D feature space. Regression lines thread
through point clouds. Decision boundaries slice the viewport. Neural
networks stack as node-and-wire diagrams. Gradient descent carves a
path down a loss landscape you can orbit. The foundations of ML for
computational designers - made visible, made understandable.
------------------------------------------------------------------------
DESCRIPTION
------------------------------------------------------------------------
Machine learning is transforming design: surrogate models replace
slow simulations, generative algorithms explore design spaces, and
Bayesian optimisation guides parametric search. But what are these
tools actually doing?
This script makes it visible.
Drop a single C# component into Grasshopper, connect a slider (1-12),
and watch ML concepts come alive as 3D geometry you can orbit, zoom,
and bake into Rhino layers.
THE 12 STEPS:
Step 1 - Data & Features
Samples scatter through 3D feature space as coloured spheres. Two
classes separate into clusters. Normalisation diagram shows raw vs
scaled data. Train/test split boxes. Distance rings show feature
distances. 187 labels - the densest step in the series.
Step 2 - Linear Regression
20 data points with a computed best-fit line (real regression, not
faked). Residual error lines connect each point to the prediction.
MSE loss box. Polynomial regression hint. Normal equation and
gradient descent formulas. Mapped to Plane.FitPlaneToPoints().
Step 3 - Classification
Two classes separated by a decision boundary line. Full sigmoid
curve plotted point-by-point. Threshold annotation at 0.5.
Multi-class softmax hint. Misclassified point marked. Linear
separability concept explained.
Step 4 - K-Nearest Neighbours
Query point surrounded by K=5 neighbourhood circle. Distance
lines to all 5 neighbours. Vote tally showing class prediction.
K choice guide (K=1 overfits, K=large underfits, use odd K).
Distance metrics comparison. Voronoi = K=1 decision regions.
Step 5 - Decision Trees
Full tree diagram: root node ("height > 10m?"), two internal
nodes, four leaf predictions (PASS/FAIL). Entropy formula with
purity explanation. Information gain. Decision regions shown as
axis-aligned splits in feature space. Random forest note.
Step 6 - Neural Networks
Full network diagram: 3 input nodes (area, height, glazing),
two hidden layers of 5 nodes each, 2 output nodes (energy, cost).
All connections drawn as wires. Forward pass and backpropagation
arrows. Activation function reference (ReLU, Sigmoid, Tanh,
Softmax). Weight annotation. Neuron equation.
Step 7 - Gradient Descent
3D loss landscape as a wireframe paraboloid surface you can orbit.
Gradient descent path of 8 steps from random start to minimum.
Learning rate comparison: too small (crawls), just right (converges),
too large (diverges). Axis labels (w1, w2, Loss). Update rule.
Batch vs SGD vs mini-batch comparison.
Step 8 - Overfitting & Regularisation
Three side-by-side diagrams: underfit (straight line), good fit
(smooth curve), overfit (wiggly line through every point). Bias-
variance tradeoff arrow from high bias to high variance. Sweet
spot marker. Regularisation methods: L1/L2, dropout, early stopping,
cross-validation.
Step 9 - Clustering (K-Means)
36 points in 3 colour-coded clusters with computed centroids.
Algorithm steps listed. Elbow method diagram with loss vs K curve
and marked elbow point. K-means++ initialisation note. Inertia
formula.
Step 10 - Dimensionality Reduction (PCA)
30-point elliptical cloud with computed mean. PC1 arrow (max
variance direction) and PC2 arrow (perpendicular, min variance).
Projection lines dropping every point onto PC1. 1D projection
shown below. Eigenvalue bar chart with explained variance
percentages. Mapped to Plane.FitPlaneToPoints() = PCA!
Step 11 - Evaluation Metrics
Large 4-cell confusion matrix with TP=42, FN=8, FP=5, TN=45.
All metrics computed from the matrix: Accuracy 87%, Precision 89%,
Recall 84%, F1 86%. Each metric annotated with plain-English
meaning. Regression metrics: MSE, RMSE, MAE, R-squared.
Precision-recall tradeoff note.
Step 12 - ML in Design
Six-stage workflow pipeline (Parameters -> Simulate -> Collect Data
-> Train Model -> Predict -> Optimise). Surrogate model diagram
with speed annotation (ms vs hours). Generative design grid of
varied forms. Bayesian optimisation acquisition function curve
with "next sample" marker. Complete tools list.
KEY FEATURES:
Single file - one C# script, paste and go
12 interactive steps - slider-driven, instant 3D diagrams
1,481 lines - comprehensive ML coverage
187 annotated labels - highest density in the entire series
8 colour-coded material branches - visual clarity
Working algorithms - real regression computed in Step 2
3D loss landscape - orbit gradient descent in Rhino
Bake to layers - organised Rhino layers with TextDot labels
Design context - every concept linked to architecture/engineering
Zero plugins - just Rhino + Grasshopper
Rhino 7 and 8 compatible (Windows and Mac)
Code IS the textbook - read the source, learn the ML
------------------------------------------------------------------------
TARGET AUDIENCE
------------------------------------------------------------------------
Computational designers adding ML to their workflow
Architecture students learning data-driven design
Grasshopper users wanting to understand surrogate models
Engineers exploring ML-guided optimisation
Plugin developers building ML-powered tools
Anyone learning ML who thinks visually
Design professionals evaluating Hops/Python ML pipelines
Researchers bridging parametric design and data science
------------------------------------------------------------------------
TAGS / KEYWORDS
------------------------------------------------------------------------
machine learning, ML, neural network, regression, classification,
KNN, decision tree, random forest, gradient descent, overfitting,
regularisation, clustering, K-means, PCA, dimensionality reduction,
confusion matrix, precision, recall, F1, surrogate model, generative
design, Bayesian optimisation, Grasshopper, Rhino, C#, RhinoCommon,
parametric, computational design, data science, study guide, 3D,
scikit-learn, Hops, Python, deep learning
------------------------------------------------------------------------
PRICING
------------------------------------------------------------------------
Standalone $14.99
Dual Bundle (ML + Linear Algebra) $24.99
Dual Bundle (ML + DSA) $24.99
Dual Bundle (ML + Comp Geometry) $24.99
Triple Bundle (ML + LinAlg + DSA) $32.99
Mega Bundle (all 11 scripts) $79.99
------------------------------------------------------------------------
SOCIAL MEDIA COPY
------------------------------------------------------------------------
TWITTER / X (280 chars):
Learn Machine Learning in Grasshopper 🧠 12 interactive 3D steps:
regression, neural nets, gradient descent, clustering, PCA,
evaluation. 187 labels, 1,481 lines. Single C# script. Zero
plugins. ML for computational designers - decoded.
INSTAGRAM / LINKEDIN:
What actually happens inside a neural network?
Data points scatter through 3D feature space. Regression lines
thread through point clouds showing residual errors. Decision
boundaries slice space into classification regions. Neural networks
stack as node-and-wire diagrams with forward pass and backprop
arrows. Gradient descent carves a path down a loss landscape you
can orbit in Rhino.
12 interactive steps. 1,481 lines. 187 annotated labels. One C#
script. Zero plugins.
Every concept mapped to a design application: surrogate models,
generative design, Bayesian optimisation. The densest product in
the series.
Link in bio. #MachineLearning #Grasshopper #Rhino3D
#ComputationalDesign #NeuralNetwork #DataScience #AI
------------------------------------------------------------------------
THUMBNAIL SUGGESTIONS
------------------------------------------------------------------------
1. Step 7 - 3D loss landscape with gradient descent path (most dramatic)
2. Step 6 - Neural network node-and-wire diagram
3. Step 1 - 3D feature space with coloured class clusters
4. Step 11 - Confusion matrix with computed metrics
------------------------------------------------------------------------
CROSS-SELL SUGGESTIONS
------------------------------------------------------------------------
Learn Linear Algebra ($14.99) - vectors, matrices, transforms,
the math behind every weight matrix and gradient computation
Learn DSA ($14.99) - data structures powering ML pipelines
(arrays, trees, hash maps, graphs)
Learn Computational Geometry ($14.99) - spatial algorithms that
combine with ML for geometric deep learning
Learn Physics Simulation ($14.99) - the simulations ML surrogates
replace (Kangaroo, particles, form-finding)
Learn Structural Analysis ($14.99) - engineering simulations
that benefit from ML surrogate modelling
Learn Design Patterns ($14.99) - software architecture for
building clean ML pipelines
Learn C# ($14.99) - language fundamentals for RhinoCommon
Learn Mesh Geometry ($14.99) - mesh data that feeds geometric ML
Learn Python ($14.99) - the language of scikit-learn and PyTorch
Learn Linux ($14.99) - server environments for ML training
Chateau Elegance v11 ($29.99) - production Grasshopper tool
------------------------------------------------------------------------
2026 Wickerson Studios - wickersonstudios.com
------------------------------------------------------------------------