apache/beam · error · ValueError

"IQR.learn_one expected univariate input, but got

Error message

"IQR.learn_one expected univariate input, but got %s", str(x)

What it means

IQR.learn_one updates the Q1/Q3 quantile trackers and expects x to be a beam.Row with exactly one field (univariate). If the row has zero or multiple fields, a ValueError is raised. This guards the IQR detector, which mathematically only operates on a single value per data point.

Solutions

  1. Project the input to a single numeric field, e.g. rows | beam.Map(lambda r: beam.Row(value=r.value)).
  2. Split multi-column data into separate detector pipelines, one per metric.
  3. Use a multivariate-capable detector if the input is intentionally multi-dimensional.

Example fix

# before
pc | anomaly.IQR(overrides).learn_one()  # Row(value=1.0, label='x')
# after
pc | beam.Map(lambda r: beam.Row(value=r.value)) | anomaly.IQR(overrides).learn_one()
Defensive patterns

Strategy: validation

Validate before calling

if len(row.__dict__) != 1:
    raise ValueError(f"IQR.learn_one requires exactly one field, got: {list(row.__dict__)}")

Type guard

def is_univariate(row): return len(getattr(row, '__dict__', {})) == 1

Try / catch

try:
    detector.learn_one(row)
except ValueError as e:
    logger.warning("skipping non-univariate row %r: %s", row, e)

Prevention

When it happens

Trigger: Passing a beam.Row with more than one attribute (multivariate input) or an empty row to IQR.learn_one, e.g. beam.Row(value=..., timestamp=...) fed directly into the detector's learn step.

Common situations: Users convert a multi-column PCollection into a Row without projecting down to one numeric column; a pipeline schema change adds a second field to the Row; accidentally passing a named tuple or dict-backed Row with metadata fields.

Understand the failure class

Background: "Must be a positive integer", "Invalid value", "Unsupported": the invalid-argument-value error family, when a library rejects the value you pass — this error's family across 35 libraries.

Related errors

AI-assisted analysis of apache/beam@12126d8942 (2026-09-13). Data as JSON: /api/errors/ca85cdd31c8d51f2. Report an issue: GitHub.

Appendix: source

Thrown at sdks/python/apache_beam/ml/anomaly/detectors/iqr.py:87

    self._q1_tracker = q1_tracker or \
        BufferedSlidingQuantileTracker(DEFAULT_WINDOW_SIZE, 0.25)
    assert self._q1_tracker._q == 0.25, \
        "q1_tracker must be initialized with q = 0.25"

    self._q3_tracker = q3_tracker or \
        SecondaryBufferedQuantileTracker(self._q1_tracker, 0.75)
    assert self._q3_tracker._q == 0.75, \
        "q3_tracker must be initialized with q = 0.75"

  def learn_one(self, x: beam.Row) -> None:
    """Updates the quantile trackers with a new data point.

    Args:
      x: A `beam.Row` containing a single numerical value.
    """
    if len(x.__dict__) != 1:
      raise ValueError(
          "IQR.learn_one expected univariate input, but got %s", str(x))

    v = next(iter(x))
    self._q1_tracker.push(v)
    self._q3_tracker.push(v)

  def score_one(self, x: beam.Row) -> Optional[float]:
    """Scores a data point based on its deviation from the IQR.

    Args:
      x: A `beam.Row` containing a single numerical value.

    Returns:
      float | None: The anomaly score.
    """
    if len(x.__dict__) != 1:
      raise ValueError(
          "IQR.score_one expected univariate input, but got %s", str(x))

View on GitHub (pinned to 12126d8942)