Description

n8n-nodes-sklearn

Custom n8n community nodes for integrating scikit-learn machine learning algorithms into your n8n workflows.

Important: Self-Hosted n8n Only

This package requires Python with scikit-learn installed and uses child_process to execute Python scripts. It is designed for self-hosted n8n installations only and is not compatible with n8n Cloud.

Available Nodes (40)

Regression

• Linear Regression – Simple and multiple linear regression
• Ridge / Lasso – L2 and L1 regularized regression
• Elastic Net – Combined L1+L2 regularization
• SVR – Support Vector Regression
• Decision Tree Regressor – Tree-based regression
• Random Forest Regressor – Ensemble tree regression
• Gradient Boosting Regressor – Boosted tree regression
• KNN Regressor – K-Nearest Neighbors regression
• MLP Regressor – Neural network regression

Classification

• Logistic Regression – Binary and multiclass classification
• SVC – Support Vector Classification
• Decision Tree Classifier – Tree-based classification
• Random Forest Classifier – Ensemble tree classification
• Gradient Boosting Classifier – Boosted tree classification
• KNN Classifier – K-Nearest Neighbors classification
• Naive Bayes – Gaussian, Multinomial, Bernoulli variants
• MLP Classifier – Neural network classification

Clustering

• KMeans – Centroid-based clustering
• DBSCAN – Density-based clustering
• Agglomerative Clustering – Hierarchical clustering
• Spectral Clustering – Graph-based clustering
• Mean Shift – Mode-seeking clustering

Anomaly Detection

• Isolation Forest – Outlier detection

Preprocessing

• Standard Scaler – Zero mean, unit variance scaling
• MinMax Scaler – Scale to [0,1] range
• Robust Scaler – Outlier-robust scaling (median/IQR)
• Normalizer – Row-wise L1/L2 normalization
• Binarizer – Threshold-based binarization
• Label Encoder – Encode categorical labels
• One Hot Encoder – One-hot encode categories
• Simple Imputer – Handle missing values

Feature Engineering

• PCA – Principal Component Analysis
• Truncated SVD – Dimensionality reduction for sparse data
• NMF – Non-negative Matrix Factorization
• Polynomial Features – Generate polynomial features
• TF-IDF Vectorizer – Text to TF-IDF features
• Feature Selection – SelectKBest, RFE, Variance Threshold

Ensemble Methods

• Voting Classifier – Combine multiple classifiers
• Stacking Classifier – Stacked generalization

Calibration

• Calibrated Classifier CV – Probability calibration

Utilities

• Train Test Split – Split data for training/testing
• Metrics – Classification, regression, clustering metrics
• Cross Validation – K-Fold, Stratified, LOO, Shuffle Split
• Grid Search CV – Hyperparameter tuning
• Pipeline – Chain transformers and estimators
• Datasets – Load sample sklearn datasets

Requirements

Self-Hosted n8n

• n8n version 0.190.0 or higher
• Python 3.7+ installed on the server
• scikit-learn and numpy Python packages

pip install scikit-learn numpy

Why Self-Hosted Only?

This package executes Python code via child_process.spawn() to run scikit-learn algorithms. This approach:

• Requires Python runtime on the host system
• Uses Node.js child_process module (restricted on n8n Cloud)
• Cannot pass n8n's community node verification for Cloud deployment

There is no pure JavaScript implementation of scikit-learn, making this architecture necessary.

Installation

Option 1: Install from npm

cd ~/.n8n/custom
npm install n8n-nodes-sklearn

Then restart n8n.

Option 2: Install via n8n UI

1. Go to Settings > Community Nodes
2. Enter n8n-nodes-sklearn
3. Click Install
4. Restart n8n

Option 3: Docker

If using Docker, ensure Python and scikit-learn are installed in your container:

FROM n8nio/n8n:latest

USER root
RUN apk add --no-cache python3 py3-pip
RUN pip3 install scikit-learn numpy
USER node

RUN cd /home/node/.n8n/custom && npm install n8n-nodes-sklearn

Usage

Basic Workflow Example

1. Load Data – Use HTTP Request, Read CSV, or other data source
2. Preprocess – Scale features with Standard Scaler
3. Train – Train a model (e.g., Random Forest Classifier)
4. Predict – Use the trained model on new data
5. Evaluate – Calculate metrics

Train a Model

Input Data:
[
  { "feature1": 1.0, "feature2": 2.0, "target": 0 },
  { "feature1": 2.0, "feature2": 3.0, "target": 1 },
  ...
]

Parameters:
- Feature Columns: feature1,feature2
- Target Column: target
- Python Path: python3

Output:
{
  "model": "{...serialized model...}",
  "score": 0.95,
  "classes": [0, 1],
  ...
}

Make Predictions

Parameters:
- Model Data: {{ $json.model }}
- Feature Columns: feature1,feature2

Output:
{
  "feature1": 1.5,
  "feature2": 2.5,
  "prediction": 1,
  "probabilities": [0.2, 0.8]
}

Python Path Configuration

By default, nodes use python3. To specify a different Python:

1. Set the Python Path parameter in each node, or
2. Set environment variable before starting n8n:

   export PYTHON_PATH=/usr/local/bin/python3.10
   n8n start
   

Troubleshooting

Python script failed

• Verify Python 3.7+ is installed: python3 --version
• Verify scikit-learn is installed: python3 -c "import sklearn; print(sklearn.__version__)"
• Check the Python Path parameter matches your installation

Feature column not found

• Column names are case-sensitive
• Check for extra spaces in column names
• Verify the column exists in your input data

Memory issues

• Process data in smaller batches
• Increase Node.js memory: NODE_OPTIONS=--max-old-space-size=4096 n8n start

Model serialization issues

• Models are serialized using Python's pickle + base64
• Ensure the same Python/sklearn version for train and predict

Development

# Clone the repository
git clone https://github.com/arturovaine/n8n-nodes-sklearn.git
cd n8n-nodes-sklearn

# Install dependencies
npm install

# Build
npm run build

# Lint
npm run lint

# Link for local development
npm link
cd ~/.n8n/custom
npm link n8n-nodes-sklearn

License

MIT

Support

• GitHub Issues: https://github.com/arturovaine/n8n-nodes-sklearn/issues
• n8n Community Forum: https://community.n8n.io/

Acknowledgments

• n8n – Workflow automation platform
• scikit-learn – Machine learning library for Python