wavqwise 0.1.3

Pluggable temporal intelligence, Any model, Any signal, Five lines to forecast.

Quick Start | 33+ Models | Auto GPU | Real Data Demos | Colab Notebooks | Ecosystem

---

Why WavqWise?

Without WavqWise	With WavqWise
Import statsmodels for ARIMA	model="arima"
Import xgboost for ML	model="xgboost"
Import chronos for foundation	model="chronos"
Write preprocessing pipeline	Auto-handled
Write evaluation loop	Built-in
Manual GPU setup	Auto-detected
Full retrain on new data	.update() incremental
~50 lines per model	5 lines, any model

---

Quick Start

pip install wavqwise

from wavqwise import WavqPipeline

pipeline = WavqPipeline()
pipeline.load("sales.csv", target="revenue", time="date")
forecast = pipeline.forecast(horizon=30, model="arima")
forecast.plot()

That's it. No model imports. No boilerplate. Change "arima" to "xgboost" or "chronos" — same code.

---

One Import. Any Model.

from wavqwise import WavqPipeline  # The ONLY import you need

# Traditional
forecast = pipeline.forecast(horizon=30, model="moving_average")
forecast = pipeline.forecast(horizon=30, model="arima")
forecast = pipeline.forecast(horizon=30, model="ets")
forecast = pipeline.forecast(horizon=30, model="holtwinters")
forecast = pipeline.forecast(horizon=30, model="theta")

# ML
forecast = pipeline.forecast(horizon=30, model="xgboost")
forecast = pipeline.forecast(horizon=30, model="lightgbm")
forecast = pipeline.forecast(horizon=30, model="random_forest")

# Foundation Models
forecast = pipeline.forecast(horizon=30, model="chronos")
forecast = pipeline.forecast(horizon=30, model="timesfm")

# Cloud / LLM
forecast = pipeline.forecast(horizon=30, model="ollama:llama3")

# Auto-select best model
forecast = pipeline.forecast(horizon=30, model="auto")

# Ensemble
forecast = pipeline.forecast(horizon=30, model=["arima", "ets", "xgboost"])

33+ Pluggable Models

Category	Models
Traditional	Moving Average, EMA, ARIMA, SARIMA, ETS, Holt-Winters, Theta, CES, Croston, Naive, Seasonal Naive
ML	XGBoost, LightGBM, CatBoost, Random Forest, Ridge, Lasso, ElasticNet
Neural	NeuralProphet, N-BEATS, TFT
Foundation	Chronos, TimesFM, Lag-Llama, Moirai, HuggingFace Hub
Cloud	TimeGPT, Ollama, OpenAI, Anthropic

---

Incremental Training

Most libraries retrain from scratch. WavqWise updates in-place:

pipeline = WavqPipeline()
pipeline.load(historical_data, target="sales", time="date")
pipeline.forecast(horizon=30, model="arima")

# New data arrives — update, don't retrain
pipeline.update(new_week_data)
forecast = pipeline.forecast(horizon=30)

---

Auto GPU / ONNX / TensorRT

WavqWise auto-detects the best available hardware on startup:

pipeline = WavqPipeline()              # Auto-detect
pipeline = WavqPipeline(device="cuda")  # Force CUDA
pipeline = WavqPipeline(device="tensorrt")  # Force TensorRT

print(pipeline.runtime_info())

WavqWise Runtime Engine
========================================
Backend:  tensorrt
Device:   NVIDIA RTX 4090
GPU:      Yes
VRAM:     24.0 GB
Compute:  SM 8.9
Available: cpu, onnx-cpu, onnx-gpu, cuda, tensorrt

Detection priority: TensorRT → ONNX GPU → PyTorch CUDA → PyTorch MPS (Apple) → ONNX CPU → CPU

ONNX Export for Production

from wavqwise.runtime import ONNXExporter, ONNXPredictor

# Export trained model to ONNX
exporter = ONNXExporter()
exporter.export_sklearn(trained_model, "model.onnx", n_features=13)

# Optimize with TensorRT (FP16)
exporter.optimize_for_tensorrt("model.onnx")

# Fast inference
predictor = ONNXPredictor("model.onnx")  # Auto GPU
result = predictor.predict(input_array)
print(predictor.benchmark(input_array))   # Latency report

---

Anomaly Detection

from wavqwise import AnomalyPipeline

detector = AnomalyPipeline()
detector.load("sensor_data.csv", target="temperature", time="timestamp")
anomalies = detector.detect(method="zscore")  # or "iqr", "isolation_forest"
anomalies.plot(show_severity=True)
print(anomalies.summary())
# Anomalies: 93/10000 (0.9%) | Method: zscore

---

EEG & Signal Processing

from wavqwise import SignalPipeline

sig = SignalPipeline()
sig.load("eeg.csv", channels=["Fp1", "Fp2", "C3", "C4"], sample_rate=256)
sig.filter(low=1, high=50, notch=50)
bands = sig.extract_bands(["delta", "theta", "alpha", "beta", "gamma"])
bands.plot_bands()
events = sig.detect_events(threshold=3.0)

---

Trading & Financial Analysis

from wavqwise import WavqPipeline
from wavqwise.trading.indicators.momentum import RSIIndicator
from wavqwise.trading.indicators.trend import MACDIndicator, SMAIndicator
from wavqwise.trading.indicators.volatility import BollingerBandsIndicator

# Load real stock data
import yfinance as yf
stock = yf.download("AAPL", period="2y", auto_adjust=True).reset_index()

# Add indicators
stock = RSIIndicator(14).compute(stock)
stock = MACDIndicator().compute(stock)
stock = BollingerBandsIndicator(20, 2).compute(stock)

# Forecast
pipeline = WavqPipeline()
pipeline.load(stock, target="Close", time="Date")
forecast = pipeline.forecast(horizon=30, model="ema")

---

Database Support

# PostgreSQL / TimescaleDB
pipeline.load("postgresql://user:pass@host:5432/db",
              query="SELECT timestamp, value FROM sensors")

# SQLite
pipeline.load("sqlite:///local.db", table="readings")

# MongoDB / InfluxDB
pipeline.load("mongodb://host:27017/db", collection="data")

---

Real Data Demos

All demos use real open-source data — no synthetic:

Demo	Data Source	Run
EEG Classification	MNE Sample Dataset (real clinical EEG)	python demos/demo_eeg_real_data.py
Trading Forecast	Yahoo Finance via yfinance (AAPL)	python demos/demo_trading_real_data.py
Anomaly Detection	Real sensor readings	python demos/demo_anomaly_detection.py
Forecasting	Sales data	python demos/demo_forecasting.py
EEG Band Analysis	Signal processing	python demos/demo_eeg_analysis.py

---

Colab Notebooks

Run in browser, zero setup:

Notebook	Data	Open
EEG Classification	MNE real EEG
Trading Forecast	yfinance AAPL/TSLA/MSFT

---

CLI

# Forecast
wavqwise forecast --input data.csv --target sales --model arima --horizon 30

# Anomaly detection
wavqwise detect --input sensor.csv --target temperature --method zscore

# List models
wavqwise models

---

Install Options

pip install wavqwise                    # Core (numpy, pandas, sklearn)
pip install wavqwise[traditional]       # + ARIMA, SARIMA, ETS, Theta
pip install wavqwise[ml]               # + XGBoost, LightGBM, CatBoost
pip install wavqwise[neural]           # + NeuralProphet, N-BEATS, TFT
pip install wavqwise[foundation]       # + Chronos, TimesFM, Moirai
pip install wavqwise[signals]          # + EEG, spectral analysis (MNE)
pip install wavqwise[trading]          # + yfinance, indicators, backtest
pip install wavqwise[database]         # + PostgreSQL, MongoDB, InfluxDB
pip install wavqwise[onnx-gpu]         # + ONNX Runtime GPU
pip install wavqwise[tensorrt]         # + TensorRT optimization
pip install wavqwise[all]              # Everything

---

Docker

docker-compose -f docker/docker-compose.yml up -d
# WavqWise + TimescaleDB + Grafana ready at localhost:8888

Ecosystem

WavqWise is part of the VK-Ant pluggable AI ecosystem — six libraries, one architecture:

Library	Domain	Tagline	PyPI
SightRAG	Vision	See. Search. Retrieve.
Sonarwise	Audio	Hear. Search. Retrieve.
Docqwise	Documents	Read. Extract. Retrieve.
WavqWise	Temporal	Sense. Forecast. Alert.
Adaptive Intelligence	Orchestration	Learn. Remember. Adapt.
LLMEvalKit	Evaluation	Evaluate. Score. Improve.

---

Contributing

git clone https://github.com/VK-Ant/wavqwise.git
cd wavqwise
pip install -e ".[dev]"
make test          # 18 tests (smoke + sanity + A/B)
make lint          # ruff + mypy

---

License

Apache 2.0 License

Author

Venkatkumar Rajan