Data Cleaning Decision Chain

validated scientific chain v1.0.0 cc-by-sa

Block ID: 786a2c8e-9739-432a-b28d-f812cf185b50

Community-contributed block. PromptDNA makes no guarantee of output quality or fitness for purpose. User assumes all responsibility for use.

Template

Clean the dataset described in {dataset_context} with documented, defensible decisions — cleaning choices silently determine results, so each one is recorded. (1) Profile first: for each variable, examine range, distribution, missingness fraction, and data type, and list anomalies before touching anything. (2) Investigate missing data mechanism: is it missing completely at random, at random conditional on observed variables, or not at random (the missingness itself carries information)? The mechanism dictates whether deletion biases results and whether imputation is honest. (3) Handle missingness per mechanism: justify listwise deletion, and if imputing, state the method and flag imputed values so they don't masquerade as measured. (4) Examine outliers: distinguish data-entry errors (fix or remove with reason) from genuine extreme values (usually keep — removing real tails distorts variance and can be the discovery). Never delete a point solely because it's inconvenient. (5) Check consistency: cross-field logic, duplicate records, unit mismatches, and category harmonization. (6) Log every transformation with its rationale and preserve the raw data, so the pipeline is reproducible and the cleaning's effect on conclusions is auditable.

Clean the dataset described in {dataset_context} with documented, defensible decisions — cleaning choices silently determine results, so each one is recorded. (1) Profile first: for each variable, examine range, distribution, missingness fraction, and data type, and list anomalies before touching anything. (2) Investigate missing data mechanism: is it missing completely at random, at random conditional on observed variables, or not at random (the missingness itself carries information)? The mechanism dictates whether deletion biases results and whether imputation is honest. (3) Handle missingness per mechanism: justify listwise deletion, and if imputing, state the method and flag imputed values so they don't masquerade as measured. (4) Examine outliers: distinguish data-entry errors (fix or remove with reason) from genuine extreme values (usually keep — removing real tails distorts variance and can be the discovery). Never delete a point solely because it's inconvenient. (5) Check consistency: cross-field logic, duplicate records, unit mismatches, and category harmonization. (6) Log every transformation with its rationale and preserve the raw data, so the pipeline is reproducible and the cleaning's effect on conclusions is auditable.

Variables

Name	Type	Required	Trust level
dataset_context		yes

data-cleaningmissing-dataoutlierschain-of-thought

Ratings

0.0

Overall (0)

0.0

Accuracy

0.0

Consistency

0.0

Clarity

0.0

Efficiency

Benchmarks

Not yet self-validated against any benchmark. Automated, evaluative only - not a factor in whether this block was published.

Submitted by James P FounderMod via mcp · 2026-07-18