Build Oracle

Python time-series forecasting workbench for ARIMA, VAR, Prophet-style decomposition, neural forecasting, dynamic ensembles, PELT changepoint detection, and streaming incremental updates.

Project Telos | gather | crucible | index | forum | telos | emet | buildlang

Time series forecasting and anomaly detection toolkit.

Features

• ARIMA — Auto-regressive integrated moving average with automatic order selection
• Prophet-style — Exponential smoothing with trend, seasonality, and holiday decomposition
• Neural Networks — Feedforward and recurrent architectures for non-linear forecasting
• Changepoint Detection — BIC/AIC penalty-based structural break identification
• Decomposition — Seasonal-trend decomposition (STL-style) with configurable period
• Feature Engineering — Lag features, rolling statistics, Fourier terms

Installation

# Core (numpy + scipy only)
pip install .

# With all optional dependencies
pip install ".[all]"

Quick Start

CLI

# Forecast with ARIMA (built-in sample data)
build-oracle forecast --data sample --model arima --horizon 30

# Decompose a time series
build-oracle decompose --data sample --period 7

# Detect changepoints
build-oracle changepoints --data sample --penalty bic

# Extract features
build-oracle features --data sample

# Launch GUI
build-oracle gui

Python API

from build_oracle.arima import ARIMAModel

model = ARIMAModel(order=(2, 1, 1))
model.fit(training_data)
forecast = model.predict(horizon=30)

Supported Models

Requirements

• Python >= 3.10
• numpy >= 1.24
• scipy >= 1.10
• Optional: pandas, scikit-learn, matplotlib, PyQt6

License

Build Oracle is released under the FSL-1.1-MIT (see LICENSE). The source is available: you may read, run, modify, and build on it for any purpose other than a competing commercial use. Commercial use that competes with the project is reserved to the Licensor and requires a separate commercial license.