> ## Documentation Index > Fetch the complete documentation index at: https://docs.nimbusbci.com/llms.txt > Use this file to discover all available pages before exploring further. # Python SDK > Overview of the nimbus-bci Python SDK: sklearn-compatible Bayesian classifiers with MNE integration for motor imagery, P300, and SSVEP. # Python SDK The **nimbus-bci** Python library provides sklearn-compatible Bayesian classifiers for Brain-Computer Interface applications. It delivers local probabilistic inference with uncertainty quantification, streaming support, and rich diagnostics. Optional extras add `NimbusSoftmax` via JAX (`pip install nimbus-bci[softmax]`) and pyRiemann feature pipelines (`pip install nimbus-bci[riemann]`). ## Start Here Install and verify your Python SDK environment. Build your first Bayesian BCI classifier in minutes. Compare NimbusLDA, NimbusQDA, NimbusSoftmax, and NimbusSTS. ## Overview `nimbus-bci` is a production-ready Python package available on PyPI that brings Bayesian inference to BCI applications with a familiar sklearn-compatible API. ```bash theme={null} pip install nimbus-bci ``` **New in `nimbus-bci`:** `CalibrationSession` packages active-learning helpers into a first-class workflow, and `nimbus_bci.riemann.make_riemann_nimbus_pipeline(...)` composes pyRiemann covariance/tangent features with Nimbus heads. See [Active Learning](/python-sdk/active-learning) and [API Reference](/python-sdk/api-reference). ## Key Features Drop-in replacement for sklearn classifiers. Works with pipelines, cross-validation, and GridSearchCV. Real-time chunk-by-chunk processing for online BCI applications with weighted aggregation. Seamless integration with MNE-Python for EEG preprocessing and feature extraction. Optional pyRiemann covariance/tangent feature pipelines feeding existing Nimbus classifier heads. Entropy, calibration metrics (ECE/MCE), ITR calculation, and quality assessment. Update models incrementally with `partial_fit()` without retraining from scratch. Rank unlabeled trials, request labels only when useful, and stop calibration when the posterior stabilizes. No API keys required. All processing happens on your machine for maximum privacy. ## Quick Example ```python theme={null} from nimbus_bci import NimbusLDA import numpy as np # Create and fit classifier clf = NimbusLDA() clf.fit(X_train, y_train) # Predict with uncertainty predictions = clf.predict(X_test) probabilities = clf.predict_proba(X_test) # Works with sklearn pipelines from sklearn.pipeline import make_pipeline from sklearn.preprocessing import StandardScaler pipe = make_pipeline(StandardScaler(), NimbusLDA()) pipe.fit(X_train, y_train) ``` ## Available Classifiers The Python SDK provides four Bayesian classifiers: | Classifier | Description | Best For | | ----------------- | -------------------------------------------------------- | -------------------------------------------- | | **NimbusLDA** | Bayesian LDA with shared covariance | Fast, well-separated classes (motor imagery) | | **NimbusQDA** | Bayesian QDA with class-specific covariances | Complex distributions (P300) | | **NimbusSoftmax** | Bayesian logistic regression (Polya-Gamma VI) | Non-Gaussian decision boundaries | | **NimbusSTS** | Bayesian Structural Time Series (Extended Kalman Filter) | Non-stationary data, long sessions | All classifiers provide: * Full posterior distributions * Uncertainty quantification * Online learning via `partial_fit()` * sklearn-compatible API **Note**: NimbusSTS also includes state management methods (`propagate_state()`, `get_latent_state()`, `set_latent_state()`) for temporal adaptation. ## Python vs Julia SDK Both SDKs implement the same Bayesian models but with different design philosophies: | Feature | Python SDK | Julia SDK | | ---------------- | ------------------------------------------------------ | -------------------------------- | | **Installation** | `pip install nimbus-bci` | `Pkg.add("NimbusSDK")` + API key | | **API Style** | sklearn-compatible classes | Functional + types | | **Integration** | sklearn pipelines, MNE-Python | RxInfer.jl ecosystem | | **Performance** | Fast local Python implementation; optional JAX Softmax | Fastest (native Julia) | | **API Key** | Not required | Required | | **Use Case** | Production ML pipelines | Research, custom models | | **License** | Proprietary (free evaluation) | Commercial tiers | ### When to Use Python SDK Choose the Python SDK if you:
✅ Work primarily in Python
✅ Use sklearn pipelines and MNE-Python
✅ Want immediate installation without API keys
✅ Need sklearn-compatible API for easy integration
✅ Prefer fully local processing ### When to Use Julia SDK Choose the Julia SDK if you:
✅ Need maximum performance (native Julia)
✅ Want to build custom Bayesian models
✅ Work with RxInfer.jl ecosystem
✅ Need pre-trained model distribution
✅ Require enterprise support ## Architecture ``` EEG Data → MNE-Python → nimbus-bci → Your Application (Preprocessing) (Classification) (Control/Analysis) ``` **Key Design Principles:** * **Local Processing**: All computation happens on your machine * **Privacy First**: Your EEG data never leaves your computer * **sklearn Compatible**: Works with existing ML workflows * **Fast Inference**: Efficient local inference, with optional JAX support for `NimbusSoftmax` * **Uncertainty Aware**: Full Bayesian posterior distributions ## Use Cases Control devices with imagined movements using CSP features and NimbusLDA Communication interfaces using ERP features and NimbusQDA Streaming inference for live brain state monitoring sklearn-compatible API for reproducible BCI research ## What's Included The `nimbus-bci` package provides: **Classifiers:** * `NimbusLDA` - Bayesian Linear Discriminant Analysis * `NimbusQDA` - Bayesian Quadratic Discriminant Analysis * `NimbusSoftmax` - Bayesian Multinomial Logistic Regression * `NimbusSTS` - Bayesian Structural Time Series (Extended Kalman Filter) **Data Structures:** * `BCIData` - Container for features, metadata, labels * `BCIMetadata` - Metadata for BCI experiments **Inference:** * `predict_batch()` - Batch inference with diagnostics * `StreamingSession` - Real-time chunk processing **Active Learning:** * `CalibrationSession` - Manage pool-based calibration rounds and stopping snapshots * `suggest_next_trial()` - Rank unlabeled feature pools by informativeness * `should_query()` - Decide whether an arriving trial is worth labeling * `calibration_sufficient()` - Stop calibration when the posterior stabilizes **Metrics:** * `compute_entropy()` - Prediction uncertainty * `compute_calibration_metrics()` - ECE/MCE * `calculate_itr()` - Information Transfer Rate * `assess_trial_quality()` - Quality checks **Utilities:** * `estimate_normalization_params()` - Feature normalization * `diagnose_preprocessing()` - Preprocessing diagnostics * `compute_fisher_score()` - Feature discriminability **MNE Integration:** * `from_mne_epochs()` - Convert MNE Epochs to BCIData * `extract_csp_features()` - CSP feature extraction * `extract_bandpower_features()` - Bandpower features **Optional Riemannian Features:** * `make_riemann_nimbus_pipeline()` - Build an EEG-epoch pipeline from pyRiemann covariance/tangent transforms to a Nimbus classifier head ## Next Read Install nimbus-bci and verify your setup Build your first BCI classifier in 5 minutes Complete API documentation with examples Reduce calibration time with label-efficient trial selection Use nimbus-bci with MNE-Python ## Support Need help? We're here for you: * **Email**: [hello@nimbusbci.com](mailto:hello@nimbusbci.com) * **Documentation**: Browse our comprehensive guides * **GitHub**: [github.com/nimbusbci/nimbuspysdk](https://github.com/nimbusbci/nimbuspysdk) * **Examples**: Check out working code samples in the package *** Ready to get started? Head to the [Installation Guide](/python-sdk/installation) to install nimbus-bci.