Validate a merged dataset against its source datasets

Audit a merged dataset against the two source datasets used to create it. This function is designed to catch common merge problems before analysis, including incompatible key types, duplicate key combinations, unexpected overlapping variables, unresolved .x / .y variables, and value conflicts between duplicated variables.

Usage

ValidateMerge(LeftData, RightData, MergedData, Keys)

Arguments

LeftData: A data frame used as one source for the merge.
RightData: A data frame used as the other source for the merge.
MergedData: The merged data frame to audit.
Keys: Character vector of key variables intended to define the merge. Multiple keys are supported, such as c("study_id", "TimePoint").

Value

A list with merge validation results, including:

SummaryText: A plain-text summary of the merge audit.
ReadyForAnalysis: Logical value indicating whether major merge-integrity issues were absent.
Summary: One-row tibble with core merge metrics.
Fingerprint: Tibble comparing rows, columns, and unique key combinations across datasets.
KeyTypes: Tibble showing key variable classes before coercion.
Checks: Tibble summarizing validation checks and pass/warning/fail status.
SuggestedActions: Tibble with suggested review steps.
Relationship: Tibble describing key uniqueness in LeftData, RightData, and MergedData.
IDCoverage: List containing Matching, LeftOnly, and RightOnly key combinations.
DuplicateKeys: List containing duplicated key rows from Left, Right, and Merged datasets.
OverlappingVariables: Tibble of variables present in both source datasets but not listed as keys.
PotentialMergeRisk: Tibble of overlap variables that could have affected an unspecified dplyr join.
JoinAudit: Tibble showing variables present in both source datasets and whether they were specified as merge keys.
OverlapAudit: Tibble auditing all source variables by source presence and key status.
UnresolvedDuplicateVariables: Tibble of .x / .y duplicate variable pairs still present in MergedData.
DuplicateVariables: Tibble with agreement, conflict counts, and classes for duplicated variables.
SuspiciousConflicts: Subset of duplicated variables with low agreement or mismatched classes.
VariableConflicts: Long-format tibble of record-level value conflicts.

Details

This function does not perform a merge. It assumes the user has already created a merged dataset and wants to check whether the merge can be trusted.