DropNans / Manipulation Layer

Remove rows containing NaN (Not a Number) values in floating-point columns. Similar to pandas' dropna() or R's na.omit() with specific handling for IEEE 754 NaN values.

Key features:

Selective column filtering
Flexible row removal criteria (any/all conditions)
Floating-point specific cleaning

Common applications:

Data cleaning for statistical analysis
Preparing complete case analysis
Removing computation artifacts
Machine learning dataset preparation
Quality control in numerical data

Note: Only considers floating-point columns as NaN is a special floating-point value. If no columns are specified, examines all floating-point columns in the dataset.

Table

Select

[column, ...]

Floating-point columns to check for NaN values. Use cases:

Specific measurement columns
Critical numerical features
Calculated fields If empty, all floating-point columns are considered. Non-floating-point columns are ignored as they cannot contain NaN values.

How

enum

Any

Strategy for row removal based on NaN presence across specified columns. Affects how aggressively rows are filtered from the dataset.

Any ~

Remove rows if any specified column contains NaN. More aggressive filtering, ensuring complete data in all specified columns. Use cases:

Complete case analysis
Critical data completeness
Statistical procedures requiring full data

All ~

Remove rows only if all specified columns contain NaN. More conservative approach, preserving rows with partial data. Suitable for:

Preserving maximum data
Handling sparse datasets
Partial information analysis