pandas "ValueError: could not convert string to float" in CI

By Kaveh Alemi·Latchkey

pandas raises "ValueError: could not convert string to float" when a column you treat as numeric holds a value it cannot parse, like an empty string, a thousands separator, or a stray label, during astype or a numeric operation.

What this error means

A data step fails with "ValueError: could not convert string to float: 'N/A'" (or similar) during astype(float), pd.to_numeric, or a numeric aggregation in CI.

ValueError: could not convert string to float: 'N/A'

Common causes

Non-numeric sentinels in a numeric column

Values like "N/A", "-", or "" sit in a column you cast to float, and pandas cannot parse them.

Locale or formatting in the raw data

Thousands separators or currency symbols make the string unparseable as a plain float.

How to fix it

Coerce with to_numeric and handle bad values

1. Use pd.to_numeric(..., errors="coerce") so unparseable values become NaN instead of raising.
2. Decide how to handle the resulting NaNs (drop, fill, or fail explicitly).
3. Re-run the numeric step.

df["amount"] = pd.to_numeric(df["amount"], errors="coerce")

Clean formatting before conversion

Strip separators and symbols, then convert, when the raw values are numeric but formatted.

df["amount"] = (df["amount"].str.replace(",", "", regex=False)).astype(float)

How to prevent it

• Validate column dtypes against a schema in CI.
• Use errors="coerce" and assert the NaN rate is acceptable.
• Normalize sentinels and formatting at ingestion.

Frequently asked questions

Related guides

References

• Python documentation
• pip documentation
• GitHub Actions documentation