timedatamodel 0.1.4

A lightweight data model for time series data with pandas, numpy, and polars support

TimeDataModel

A lightweight Python data model for time series data with native bridges to numpy, pandas, polars, and xarray.

TimeDataModel provides a structured, metadata-rich representation of time series data that lets you:

• 📐 Structure - Represent univariate, multivariate, N-dimensional, and hierarchical time series as typed Python objects;
• 🔄 Convert - Seamlessly convert between pandas DataFrames, numpy arrays, and polars DataFrames;
• 🗺️ Locate - Attach geographic coordinates and areas to time series with spatial filtering;
• 📏 Measure - Track and convert physical units with automatic dimensional validation;
• 📊 Validate - Inspect data quality with coverage bars, gap detection, and consistency checks;
• 💾 Serialize - Read and write time series to JSON and CSV with full metadata preservation.

⬇️ Installation  |  📖 Documentation  |  🚀 Examples

---

🧱 Core Data Classes

TimeDataModel represents time series data at multiple levels of complexity. The table below gives a summary of the available data classes.

Class	Description
📈 TimeSeries	Univariate time series with scalar metadata (name, unit, frequency, timezone)
📊 TimeSeriesTable	Multivariate time series — multiple columns sharing the same timestamp index
🧊 TimeSeriesArray	N-dimensional time series with named dimensions and label-based selection
📦 TimeSeriesCollection	Heterogeneous container for series with different indices
🌳 HierarchicalTimeSeries	Tree-structured time series with aggregation across hierarchy levels
🗺️ GeoLocation / GeoArea	Geographic point and polygon types with distance, bearing, and containment
⏱️ Frequency	ISO 8601 duration-based frequencies (PT1H, P1D, P1M, etc.)
🏷️ DataType	Hierarchical taxonomy: ACTUAL → OBSERVATION, DERIVED; CALCULATED → ESTIMATION, REFERENCE, etc.

---

🔌 Integrations

Each class provides to_X, from_X, and apply_X bridges to popular array and dataframe libraries.

	numpy	pandas	polars	xarray
TimeSeries
to_X	✅	✅	✅	✅
from_X	—	✅	✅	✅
apply_X	✅	✅	✅	✅
TimeSeriesTable
to_X	✅	✅	✅	✅
from_X	—	✅	—	✅
apply_X	✅	✅	✅	✅
TimeSeriesArray
to_X	✅	✅	—	✅
from_X	✅	—	—	✅
apply_X	—	✅¹	✅¹	✅

¹ Gated: raises ValueError if the array has more than 2 non-time dimensions.

---

🚀 Quick Start

from datetime import datetime, timezone
from timedatamodel import TimeSeries, Frequency

# Create a univariate time series
ts = TimeSeries(
    Frequency.PT1H,
    timezone="UTC",
    timestamps=[datetime(2024, 1, 1, i, tzinfo=timezone.utc) for i in range(24)],
    values=[100.0 + i * 2.5 for i in range(24)],
    name="power_output",
    unit="MW",
)

# Convert to pandas DataFrame
df = ts.to_pandas_dataframe()

# Convert to numpy array
arr = ts.to_numpy()

# Arithmetic operations
ts_doubled = ts * 2
ts_shifted = ts + 50

# Inspect data quality
print(ts.coverage_bar())

# Serialize to JSON
json_str = ts.to_json()

---

✨ Key Features

• 📐 Multi-dimensional - From single series (TimeSeries) to N-dimensional arrays (TimeSeriesArray) with .sel() and .isel() for label- and index-based selection;
• 🔢 Arithmetic - Element-wise +, -, *, /, comparisons, and abs() with automatic NaN handling;
• 🗺️ Geospatial - Attach locations, compute distances (Haversine), filter by radius or area, and find nearest neighbors;
• 🌳 Hierarchical - Build tree structures with automatic aggregation (SUM, MEAN, MIN, MAX) and cross-level unit conversion;
• 📏 Units - Optional pint integration for physical unit tracking and conversion;
• 📊 Data Quality - Coverage bars (terminal + HTML), missing-value detection, and timestamp validation;
• 💾 Serialization - JSON and CSV I/O with full round-trip metadata preservation;
• 🐍 Type-safe - Full type hints with PEP 561 support.

---

⬇️ Installation

Install the stable release:

pip install timedatamodel

Install with optional dependencies:

pip install timedatamodel[pandas]    # pandas support
pip install timedatamodel[polars]    # polars support
pip install timedatamodel[geo]       # geospatial support (shapely)
pip install timedatamodel[pint]      # unit handling (pint)
pip install timedatamodel[all]       # everything

Install in editable mode for development:

git clone https://github.com/rebase-energy/TimeDataModel.git
cd TimeDataModel
pip install -e .[dev]

---

📓 Examples

Explore the library through the example notebooks:

#	Notebook	Topic
01	Getting Started	Creating and inspecting time series
02	NumPy & Pandas	Converting to and from numpy and pandas
03	Unit Handling	Physical unit conversion with pint
04	Operations	Arithmetic, slicing, and indexing
05	Multivariate	Working with multi-column time series
06	Arrays & Collections	N-dimensional data and heterogeneous containers
07	Data Quality	Coverage bars and validation
08	I/O	JSON and CSV serialization
09	Geospatial	Locations, areas, and spatial filtering
10	Hierarchical	Tree structures and aggregation

---

🤝 Contributing

Contributions are welcome! Here are some ways to contribute to TimeDataModel:

• Propose new data classes or extend existing ones;
• Improve documentation or add example notebooks;
• Report bugs or suggest features via GitHub Issues.

---

📄 Licence

This project uses the MIT Licence.