dqm-ml-core 2.0.0rc1

Python library designed provide core dqml metrics without huge dependencies, as well as common API shared by metrics

DQM-ML Core

Core package for DQM-ML V2 providing the foundational API and standard metrics for data quality assessment.

Installation

pip install dqm-ml-core

Note: dqm-ml-core provides metric processors only — no CLI or job orchestration. Use directly via Python or with dqm-ml-job for YAML config execution.

Quick Start

Completeness Example

from dqm_ml_core import CompletenessProcessor

processor = CompletenessProcessor(
    name="my_check",
    config={"input_columns": ["col_a", "col_b"]}
)
result = processor.compute({})
print(f"Completeness: {result['overall_completeness']}")

Representativeness Example

from dqm_ml_core import RepresentativenessProcessor
import numpy as np

# Create sample data (e.g., 1000 samples from normal distribution)
data = np.random.randn(1000)

processor = RepresentativenessProcessor(
    name="dist_check",
    config={
        "input_columns": ["feature"],
        "distribution": "normal",
        "metrics": ["chi-square", "kolmogorov-smirnov"],
        "distribution_params": {"mean": 0.0, "std": 1.0}
    }
)

result = processor.compute({})
print(f"Chi-Square p-value: {result['feature_chi-square_pvalue']}")
print(f"KS p-value: {result['feature_kolmogorov-smirnov_pvalue']}")

With dqm-ml-job

For running from a YAML config, install together with dqm-ml-job:

pip install dqm-ml-job dqm-ml-core

Then use this config:

dataloaders:
  train:
    type: parquet
    path: data/train.parquet

metrics_processor:
  completeness:
    type: completeness
    input_columns: [col_a, col_b]
  
  representativeness:
    type: representativeness
    input_columns: [feature_x]
    distribution: "normal"

Key Concepts

DatametricProcessor

The base class for all metrics and feature extractors. It supports a streaming architecture by splitting computation into two phases:

1. Batch Level: compute_batch_metric() updates intermediate statistics for a single chunk of data.
2. Dataset Level: compute() aggregates these statistics into final scores.

Included Metrics

Metric	Description
Completeness	Analyzes null/missing values in your dataset
Representativeness	Statistical distribution analysis (Chi-Square, KS, Shannon Entropy, GRTE)

For Developers

To create a new metric:

1. Subclass dqm_ml_core.api.data_processor.DatametricProcessor.
2. Define needed_columns(), generated_features(), and generated_metrics().
3. Implement the streaming logic in compute_batch_metric() and compute().

Dependencies

DQM-ML is modular. For core metrics:

# Minimal: use as library only
pip install dqm-ml-core

# For YAML config execution
pip install dqm-ml-job dqm-ml-core

# Full stack with all metrics
pip install dqm-ml-job dqm-ml-core dqm-ml-images dqm-ml-pytorch

See Also

• Metrics Documentation
• API Reference