Crate scirs2_series

Crate scirs2_series 

Source
Expand description

§SciRS2 Series - Time Series Analysis

scirs2-series provides comprehensive time series analysis capabilities, offering decomposition, forecasting (ARIMA, VAR, Prophet-style), anomaly detection, change point detection, and advanced methods with parallel processing and streaming support.

§🎯 Key Features

  • Decomposition: STL, classical, SSA, MSTL, TBATS for trend/seasonality extraction
  • Forecasting: ARIMA, SARIMA, VAR, VECM, exponential smoothing
  • Anomaly Detection: Statistical, isolation forest, distance-based methods
  • Change Point Detection: PELT, binary segmentation, CUSUM, Bayesian online
  • Causality Testing: Granger causality, transfer entropy, convergent cross mapping
  • Clustering: Time series k-means, hierarchical, DTW-based clustering
  • State-Space Models: Kalman filtering, structural models, dynamic linear models
  • Transformations: Box-Cox, differencing, stationarity tests (ADF, KPSS)

§📦 Module Overview

SciRS2 ModulePython EquivalentDescription
decompositionstatsmodels.tsa.seasonal.STLTime series decomposition
arimastatsmodels.tsa.arima.model.ARIMAARIMA forecasting
varstatsmodels.tsa.vector_ar.var_model.VARVector autoregression
anomaly-Anomaly/outlier detection
changepointrupturesChange point detection
causalitystatsmodels.tsa.stattools.grangercausalitytestsGranger causality testing

§🚀 Quick Start

[dependencies]
scirs2-series = "0.1.0-rc.2"
use scirs2_series::decomposition::stl::{stl_decomposition, STLOptions};
use scirs2_core::ndarray::array;

// STL decomposition
let data = array![1.0, 2.0, 3.0, 4.0, 5.0, 4.0, 3.0, 2.0]; // Seasonal data
let options = STLOptions::default();
let result = stl_decomposition(&data, 4, &options).unwrap();
// result.trend, result.seasonal, result.residual

§🔒 Version: 0.1.0-rc.2 (October 03, 2025)

  • Change point detection
    • PELT (Pruned Exact Linear Time) algorithm
    • Binary segmentation
    • CUSUM methods
    • Bayesian online change point detection
    • Kernel-based change detection
  • Anomaly detection
    • Statistical process control (SPC)
    • Isolation forest for time series
    • Z-score and modified Z-score methods
    • Interquartile range (IQR) detection
    • Distance-based and prediction-based approaches
  • Automatic pattern detection
    • Period detection using ACF, FFT, and wavelets
    • Automatic seasonal decomposition with period detection
  • Advanced trend analysis
    • Non-linear trend estimation using splines
    • Cubic splines, B-splines, and P-splines
    • Robust trend filtering with confidence intervals
  • State-space models
    • Kalman filtering and smoothing
    • Structural time series models
    • Dynamic linear models
    • Unobserved components models
  • Causality testing and relationship analysis
    • Granger causality testing with F-statistics and p-values
    • Transfer entropy measures with bootstrap significance testing
    • Convergent cross mapping for nonlinear causality detection
    • Causal impact analysis for intervention assessment
  • Correlation and relationship analysis
    • Cross-correlation functions with confidence intervals
    • Dynamic time warping with multiple constraint types
    • Time-frequency analysis (STFT, CWT, Morlet wavelets)
    • Coherence analysis for frequency domain relationships
  • Time series clustering and classification
    • K-means, hierarchical, and DBSCAN clustering algorithms
    • Multiple distance measures (DTW, Euclidean, correlation-based)
    • k-NN classification with DTW and other distance functions
    • Shapelet discovery and shapelet-based classification
  • Vector autoregressive models
    • VAR model fitting and prediction
    • Impulse response functions
    • Variance decomposition
    • Granger causality testing
    • VECM for cointegrated series
    • Automatic order selection
  • ARIMA models with enhanced functionality
    • Automatic order selection with multiple criteria
    • Stepwise and grid search optimization
    • Seasonal ARIMA (SARIMA) support
    • Model diagnostics and information criteria
  • Time series transformations
    • Box-Cox transformations with automatic lambda estimation
    • Differencing and seasonal differencing
    • Stationarity tests (ADF, KPSS)
    • Normalization and scaling (Z-score, Min-Max, Robust)
    • Detrending and stationarity transformations
  • Dimensionality reduction for time series
    • Principal Component Analysis (PCA) for time series
    • Functional PCA for functional time series data
    • Dynamic Time Warping barycenter averaging
    • Symbolic approximation methods (SAX, APCA, PLA)
  • Time series regression models
    • Distributed lag models (DL) with flexible lag structures
    • Autoregressive distributed lag (ARDL) models with automatic lag selection
    • Error correction models (ECM) for cointegrated series
    • Regression with ARIMA errors for correlated residuals
  • Forecasting methods (ARIMA, exponential smoothing)
    • Automatic model selection
    • Seasonal and non-seasonal models
  • Feature extraction for time series
  • Feature selection methods for time series
    • Filter methods (correlation, variance, mutual information, statistical tests)
    • Wrapper methods (forward selection, backward elimination, recursive elimination)
    • Embedded methods (LASSO, Ridge, tree-based importance)
    • Time series specific methods (lag-based, seasonal, cross-correlation, Granger causality)
  • Environmental and climate data analysis
    • Temperature analysis (heat waves, growing degree days, climate normals)
    • Precipitation analysis (drought detection, SPI, rainfall classification)
    • Atmospheric analysis (storm detection, wind power, wind rose statistics)
    • Climate indices (SOI, NAO, PDSI)
    • Environmental stress index calculation
  • Biomedical signal processing
    • ECG analysis (R-peak detection, HRV, arrhythmia detection)
    • EEG analysis (seizure detection, frequency bands, connectivity)
    • EMG analysis (muscle activation, fatigue detection, onset detection)
    • Cross-signal synchronization and health assessment
  • IoT sensor data analysis
    • Environmental sensors (temperature, humidity, pressure, light)
    • Motion sensors (accelerometer, GPS, activity recognition)
    • Data quality assessment and sensor malfunction detection
    • Predictive maintenance and system health monitoring
  • Comprehensive visualization capabilities
    • Interactive time series plotting with zoom and pan
    • Forecasting visualization with confidence intervals
    • Decomposition result visualization (trend, seasonal, residual components)
    • Multi-series plotting and comparison
    • Seasonal pattern visualization
    • Anomaly and change point highlighting
    • Dashboard generation utilities
    • Export capabilities (PNG, SVG, HTML)
  • Quantum-inspired time series analysis
    • Quantum attention mechanisms using superposition principles
    • Variational quantum circuits for pattern recognition
    • Quantum kernel methods for similarity measures
    • Quantum annealing optimization for hyperparameter tuning
    • Quantum state representations for complex time series patterns
  • Advanced training methodologies
    • Model-Agnostic Meta-Learning (MAML) for few-shot forecasting
    • Neural Ordinary Differential Equations (NODEs) for continuous-time modeling
    • Variational autoencoders with uncertainty quantification
    • Bayesian neural networks for probabilistic forecasting
    • Gradient-based meta-learning optimization techniques
  • Out-of-core processing for massive datasets
    • Chunked processing with configurable chunk sizes and overlap
    • Memory-mapped file I/O for efficient disk access
    • Streaming statistics computation (mean, variance, quantiles)
    • Parallel processing of chunks with progress tracking
    • CSV and binary file format support
  • Distributed computing support for large-scale time series processing
    • Multi-node cluster coordination and task distribution
    • Load balancing and fault tolerance mechanisms
    • Distributed forecasting, feature extraction, and anomaly detection
    • Task dependency management and priority scheduling
    • Real-time cluster monitoring and performance metrics
  • Advanced Fusion Intelligence - Next-Generation AI Systems
    • Quantum-Neuromorphic fusion processors combining quantum and biological computing
    • Meta-learning systems that learn how to learn from time series patterns
    • Self-evolving neural architectures that redesign themselves autonomously
    • Consciousness-inspired computing with attention and self-awareness mechanisms
    • Temporal hypercomputing for multi-dimensional time analysis
    • Autonomous mathematical discovery and pattern recognition
    • Advanced-predictive analytics for impossible event prediction
    • Distributed quantum networks with planet-scale processing capabilities
  • Utility functions for time series operations

Modules§

advanced_advanced_visualization
Advanced Time Series Visualization Module
advanced_fusion_intelligence
Advanced Fusion Intelligence Modules
advanced_training
Advanced Training Methods for Time Series
anomaly
Anomaly detection algorithms for time series
arima_models
Enhanced ARIMA models with automatic order selection
biomedical
Biomedical signal processing for time series analysis
causality
Causality testing and relationship analysis for time series
change_point
Change point detection algorithms
clustering
Time series clustering and classification algorithms
correlation
Time series correlation and relationship analysis
decomposition
Time series decomposition methods
decomposition_compat
Time series decomposition methods (Compatibility layer)
detection
Time series pattern detection
diagnostics
Model diagnostics and validation tools for time series models
dimensionality_reduction
Time series dimensionality reduction methods
distributed
Distributed computing support for time series processing
enhanced_arma
Enhanced autoregressive and moving average models
ensemble_automl
Ensemble Forecasting Methods and AutoML for Time Series
environmental
Environmental and climate data analysis for time series
error
Error types for the time series module
feature_selection
Time series feature selection methods
features
Time series feature extraction modules
financial
Comprehensive Financial Analysis and Modeling
financial_advanced
Advanced Financial Time Series Analytics
forecasting
Time series forecasting methods
gpu_acceleration
GPU acceleration infrastructure for time series operations
iot_sensors
IoT sensor data analysis for time series
neural_forecasting
Neural Forecasting Models for Time Series
neuromorphic_computing
Neuromorphic Computing for Advanced Time Series Analysis
optimization
Optimization algorithms for time series models
out_of_core
Out-of-core processing for massive time series datasets
quantum_forecasting
Quantum-Inspired Time Series Forecasting
regression
Time series regression models
sarima_models
Seasonal ARIMA (SARIMA) models
state_space
State-space models for time series analysis
streaming
Real-time streaming time series analysis module
tests
Statistical tests for time series analysis
transformations
Time series transformations for preprocessing and analysis
trends
Trend estimation and filtering methods for time series
utils
Utility functions for time series analysis
validation
Validation utilities for time series module
var_models
Vector Autoregressive (VAR) models for multivariate time series
visualization
Comprehensive time series visualization module