Fix Missing Values

Runs 100% locally

Find blank cells (even hidden ones) and fill them in or remove them.

description Using survey_responses.csv from the upload screen.

Preview: survey_responses.csv

2,150 rows, 6 columns

	respondent_id	age	region	income	satisfaction	comments
0	R-1001	34	West	52000	4	great service
1	R-1002	N/A	East		3	?
2	R-1003	41	-	61000	NULL	none
3	R-1004	29	South	N/A	5	quick

Missingness profile

Rows

2,150

Cells missing

1,043

% cells missing

8.1%

Complete rows

1,388

column	dtype	missing	missing_pct	disguised	has_missing
respondent_id	object	0	0.0%	0	False
age	float64	187	8.7%	61	True
region	object	142	6.6%	142	True
income	float64	329	15.3%	118	True
satisfaction	float64	95	4.4%	40	True
comments	object	290	13.5%	290	True

Options

Strategy

layers Resolution order: per-column override → global strategy → preset. The most specific setting wins; layers it overrides are dimmed.

Preset

info Overridden by Global strategy → median (set under Advanced options). Presets apply only when global is “(use preset)”.

detect-only (standardize sentinels to NaN, no fill or drop) safe-fill (numeric → median, categorical → mode) drop-incomplete (drop any row with missing)

detect-only: replace 'N/A', '-', 'NULL', etc. with real NaN, then stop. safe-fill: also fill — numeric columns with median, others with mode. drop-incomplete: also drop every row that has any missing cell.

Advanced options

Detection

check Standardize disguised nulls to NaN

Blanks in disguise (N/A, dash, NULL) — comma-separated

N/A, n/a, NA, NULL, null, None, -, --, ?, #N/A

Text that really means “empty.” Matched case-insensitively after stripping whitespace.

Strategy override

Global strategy

median

drop_row / drop_col use the thresholds below. mean / median / interpolate are numeric only — non-numeric columns fall back to the categorical strategy.

Categorical fallback (for non-numeric columns)

mode

Drop thresholds

Row drop threshold (drop rows with ≥ this fraction missing across selected cols)

1.00

Column drop threshold (drop columns with ≥ this fraction missing)

1.00

Scope

Columns to handle (default: all)

respondent_id ✕ age ✕ region ✕ income ✕ satisfaction ✕ comments ✕

Columns to skip

Choose columns

Per-column strategy overrides (optional)

Set a different strategy for specific columns. Leave any row blank to use the global strategy.

Column	Override	Resolves to
age	(global)	median · global
region	(global)	mode · global → categorical fallback
income	(global)	median · global
satisfaction	(global)	median · global
comments	constant	constant · this column

Results

Sentinels → NaN

651

Cells filled

1,043

Rows dropped

Columns dropped

Missingness — before vs. after

column	before_missing	before_pct	strategy
respondent_id	0	0.0	—
age	187	8.7	median
region	142	6.6	mode
income	329	15.3	median
satisfaction	95	4.4	median
comments	290	13.5	constant

Audit (first 50 changes)

row	column	old_value	new_value	reason
2	age	N/A	37.0	fill: median
2	income	(blank)	54000.0	fill: median
2	comments	?	(no comment)	fill: constant
3	region	-	West	fill: mode
3	satisfaction	NULL	4.0	fill: median
4	income	N/A	54000.0	fill: median

… and 1,037 more (download the full audit below).

Handled preview (first 10 rows)

	respondent_id	age	region	income	satisfaction	comments
0	R-1001	34.0	West	52000.0	4.0	great service
1	R-1002	37.0	East	54000.0	3.0	(no comment)
2	R-1003	41.0	West	61000.0	4.0	none
3	R-1004	29.0	South	54000.0	5.0	quick

arrow_forwardMissing values handled. Next, most files need: Find Duplicates →