zedda.warnings()

zedda.warnings(
    path,
    sample_size=None,
    correlate=False,
) -> None

Print every data quality warning ranked by severity, with inline fix code and a copy-paste block at the end. Returns None.

# Arguments

Argument	Type	Default	Description
path	str / Path / DataFrame	(required)	Input dataset
sample_size	int or None	None	Sample size in rows
correlate	bool	False	Force O(N²) correlation

# Returns

None. warnings() is print-first. For programmatic access, use collect_warnings().

# Warning categories

Zedda detects warnings in these categories:

Category	Example
High nulls	A column has > 50% nulls
Constant column	A column has zero variance (every non-null value identical)
Outliers	A numeric column passes the outlier heuristic (mean > 0, val_max > 10×mean, etc.)
ID column	A column has very high cardinality relative to row count
High-cardinality string	A string column has very high cardinality (likely free-text)
Skewed	A numeric column has
High kurtosis	A numeric column has kurtosis > 7 (heavy-tailed)
Correlated	A pair of numeric columns has

The outlier heuristic in zedda._warnings.is_outlier_column:

• Numeric column
• mean > 0
• unique > 5
• val_max > 10
• val_max > 10 × mean
• column name does not contain “ratio” or “pct”
• mean ≥ 2.0
• not a small-int categorical (unique < 15 with val_min ≥ 0)
• not a 0…unique+5 bounded integer

# Severity levels

Each warning has a severity:

• critical — high-nulls, ID columns, missing-target
• warning — outliers, high-cardinality strings, skewed, high kurtosis
• info — correlated pairs, constant columns

# Example

import zedda as zd

zd.warnings("data.csv")

# See also

• collect_warnings() — programmatic variant returning list[dict].
• fix() — generate pandas code to address the warnings.
• ml_ready() — score the dataset for ML training.