onTime
AI & ML interests
Time series, forecasting, foundation models
Recent Activity
Your library to work with time series.
onTime is an open-source Python library for sensor data and time series: analysis, forecasting, and anomaly detection. It extends established libraries such as Darts with benchmarking tools for models and detectors, domain-specific components (energy consumption, district heating networks, …), and an Altair-based plotting API. Its plugin-style architecture lets organisations extend it while keeping proprietary components private.
Quick start
pip install ontime
import ontime as on
import pandas as pd
ts = on.generators.random_walk().generate(
start=pd.Timestamp("2023-01-01"), end=pd.Timestamp("2023-12-31")
)
model = on.context.common.GenericPredictor()
model.fit(ts)
forecast = model.predict(48)
On the Hub
This organization hosts the datasets, benchmarks, and evaluation results produced with onTime.
- onTime-Bench — an open time series forecasting benchmark with a public leaderboard (in preparation).
Resources
| 📖 Documentation | ontime.re |
| 🚀 Getting started | ontime.re/getting_started |
| 💻 Source code | github.com/ontime-re/ontime |
| 📦 Package | pypi.org/project/ontime |
| 🐛 Issues | GitHub issues |
Contributing
Contributions, bug reports, and dataset or benchmark suggestions are welcome. See the contribution guide to get started.
onTime is released under the Apache 2.0 license.