Tyche Transforms

This package provides the transformation functions defined in MARKETS_INFO.md for Tyche market data pipelines, operating on xarray DataArray inputs. The intent is to keep the same implementations between local development and the Chainlink adapter.

Install

uv pip install tyche-transforms

For development:

uv pip install -e ".[test]"

Usage

import pandas as pd
import xarray as xr
from tyche_transforms.transforms import date_first, runlen_lt, sum_value

times = pd.date_range("2024-01-01", periods=4)
data = xr.DataArray([15.0, 10.0, 8.0, 11.0], dims="time", coords={"time": times})
print(sum_value(data))
print(date_first(data, threshold=9.0))
print(runlen_lt(data, threshold=9.0))

GitHub Actions

This repo includes a workflow that runs tests on every push/PR and publishes to PyPI when you push a tag matching v* (for example v0.2.0). To enable publishing:

1. Add a repository secret named PYPI_API_TOKEN containing your PyPI token.
2. Push a version tag:

git tag v0.2.0
git push origin v0.2.0

Transformation Functions

DataArray -> Scalar

• SUM (0 args): sum over window -> scalar.
• AVG (0 args): average over window -> scalar.
• MAX (0 args): max over window -> scalar.
• INDEX_OF_MAX_VALUE (0 args): index of the max value -> scalar.
• MIN (0 args): min over window -> scalar.
• DATE_MAX (0 args): timestamp of the max value -> scalar (timestamp).
• DATE_FIRST:threshold (1 arg): timestamp of first value > threshold, or epoch pd.Timestamp(0) if none -> scalar (timestamp).
• RUNLEN_LT:threshold (1 arg): longest run length where value < threshold -> scalar.
• RUNLEN_GT:threshold (1 arg): longest run length where value > threshold -> scalar.

Scalar -> Scalar

• ABS (0 args): absolute value of a scalar -> scalar.
• SUBTRACT_V:threshold (1 arg): subtract a scalar to the given threshold -> scalar.