Fill Missing Values

Use the Fill Nulls node to replace missing values instead of dropping the rows. It works on both the polars and pandas engines.

You'll use: File Input → Fill Nulls → File Output.

Steps

1. File Input — select your dataset.
2. Fill Nulls — choose a strategy per group of columns:
   • Numeric: strategy: "median" (or "mean") with columns: ["age"].
   • Categorical: strategy: "constant", value: "Unknown", columns: ["region"].
   • Ordered/time-series: strategy: "ffill" (carry the last value forward) or "bfill" (carry the next value backward). Add a second Fill Nulls node when different columns need different strategies.
3. File Output — write the filled result.

Before / after

Tips

• For modeling, prefer imputing inside the model — a train node's Advanced → Preprocessing applies the same fill at predict time. See Feature Engineering.
• Want to drop instead of fill? Use Drop Nulls.

See also

• Fill Nulls · Drop Nulls
• Recipes