Health Assessment: This dataset has 2 critical findings that will cause silent data loss or incorrect calculations if not addressed. These should be fixed before using this data for any downstream purpose.
+Assumption: Column 'revenue_mixed' appears to be purely numeric.
+Reality: However, 15 str values were found among 200 non-null values (the majority type covers 92.5%). Examples of unexpected values: 'N/A', 'N/A', 'N/A', and 2 more.
+Impact: Rows with non-numeric values in 'revenue_mixed' will be silently dropped or converted to NaN during sums, averages, and other calculations, producing incorrect results without any error message.
+Recommended Fix: Review the non-numeric values in 'revenue_mixed' and decide whether they should be converted to numeric, replaced with a proper null, or moved to a separate column.
+Prevention Rule: Every value in 'revenue_mixed' should be the same type (numeric). Add a type-check validation rule at data entry or ingestion time.
+Assumption: Column 'revenue_mixed' appears to be a clean numeric column.
+Reality: However, 7.5% of values (1 distinct sentinel string) are placeholder text rather than real data: 'N/A' (15 times).
+Impact: Tools like pandas and Excel silently drop sentinel strings when computing sums or averages on 'revenue_mixed', making totals lower than expected. No error is raised, so the data loss goes unnoticed.
+Recommended Fix: Replace sentinel values in 'revenue_mixed' with proper null/blank cells so that downstream tools handle missing data correctly and row counts reflect reality.
+Prevention Rule: Never use placeholder text like 'N/A' in a numeric column. Use blank cells or a dedicated status column instead.
+| Field | Issue Type | Severity |
|---|---|---|
| revenue_mixed | +Type Inconsistency | +Critical | +
| revenue_mixed | +Sentinel Value | +Critical | +
Health Assessment: No critical issues were found, but 3 warnings indicate potential problems that could cause misinterpretation or key mismatches.
+Assumption: Column 'mostly_null' is expected to be fully populated.
+Reality: However, 412 of 500 rows (82.4%) are null or blank. The missing values are scattered across the dataset.
+Impact: With more than half its values missing, 'mostly_null' is unreliable for analysis. Aggregations will silently exclude the missing rows, and any model trained on this column will learn from a biased sample.
+Recommended Fix: Investigate why 'mostly_null' has missing values. If the blanks represent a known condition, consider a default value or a separate status column. If they are data-entry gaps, backfill from the source system.
+Prevention Rule: Add a NOT NULL or completeness check for 'mostly_null' at ingestion time. Flag any batch where null rate exceeds the historical baseline.
+Assumption: Column 'notes' is expected to be fully populated.
+Reality: However, 250 of 500 rows (50.0%) are null or blank. The missing values are scattered across the dataset.
+Impact: With more than half its values missing, 'notes' is unreliable for analysis. Aggregations will silently exclude the missing rows, and any model trained on this column will learn from a biased sample.
+Recommended Fix: Investigate why 'notes' has missing values. If the blanks represent a known condition, consider a default value or a separate status column. If they are data-entry gaps, backfill from the source system.
+Prevention Rule: Add a NOT NULL or completeness check for 'notes' at ingestion time. Flag any batch where null rate exceeds the historical baseline.
+Assumption: Column 'revenue' appears to be a unique identifier (ID column).
+Reality: However, 496 of 500 values are unique (uniqueness ratio: 99.2%), meaning some IDs appear more than once. Duplicate values include: '117.71', '210.24', '263.77', '9.52'.
+Impact: Duplicate IDs in 'revenue' cause row-level joins to fan out, producing unexpected extra rows in merged datasets. Aggregations that assume one row per ID will double-count affected records.
+Recommended Fix: Investigate the duplicate values in 'revenue' to determine whether they are true duplicates (same record entered twice) or legitimate repeats (one-to-many relationship). De-duplicate or re-model accordingly.
+Prevention Rule: If 'revenue' is meant to be a primary key, enforce a uniqueness constraint at the database or validation layer.
+Assumption: Column 'constant_col' is expected to carry meaningful, varying data.
+Reality: However, only 1 distinct value was found across 500 rows (uniqueness ratio: 0.2%). Most common: 'FIXED' (500 times).
+Impact: A near-constant column like 'constant_col' adds no analytical value. Including it in models or reports may mislead readers into thinking the field varies when it does not.
+Recommended Fix: Verify whether 'constant_col' should actually vary. If not, document it as a constant and consider removing it from analysis. If it should vary, investigate why the data is uniform.
+Prevention Rule: If 'constant_col' is supposed to carry diverse values, add a data-quality check that flags columns with fewer than 1% unique values.
+Assumption: Column 'discount_pct' is expected to be fully populated.
+Reality: However, 80 of 500 rows (16.0%) are null or blank. The missing values are scattered across the dataset.
+Impact: Missing values in 'discount_pct' will be silently excluded from calculations. If the missingness is not random, aggregations and models will be biased toward the non-missing subset.
+Recommended Fix: Investigate why 'discount_pct' has missing values. If the blanks represent a known condition, consider a default value or a separate status column. If they are data-entry gaps, backfill from the source system.
+Prevention Rule: Add a NOT NULL or completeness check for 'discount_pct' at ingestion time. Flag any batch where null rate exceeds the historical baseline.
+Assumption: Column 'is_renewal' is expected to carry meaningful, varying data.
+Reality: However, only 2 distinct values were found across 500 rows (uniqueness ratio: 0.4%). Most common: 'True' (266 times), 'False' (234 times).
+Impact: A near-constant column like 'is_renewal' adds no analytical value. Including it in models or reports may mislead readers into thinking the field varies when it does not.
+Recommended Fix: Verify whether 'is_renewal' should actually vary. If not, document it as a constant and consider removing it from analysis. If it should vary, investigate why the data is uniform.
+Prevention Rule: If 'is_renewal' is supposed to carry diverse values, add a data-quality check that flags columns with fewer than 1% unique values.
+Assumption: Column 'order_status' is expected to carry meaningful, varying data.
+Reality: However, only 4 distinct values were found across 500 rows (uniqueness ratio: 0.8%). Most common: 'Pending' (136 times), 'Closed' (131 times), 'Active' (127 times).
+Impact: A near-constant column like 'order_status' adds no analytical value. Including it in models or reports may mislead readers into thinking the field varies when it does not.
+Recommended Fix: Verify whether 'order_status' should actually vary. If not, document it as a constant and consider removing it from analysis. If it should vary, investigate why the data is uniform.
+Prevention Rule: If 'order_status' is supposed to carry diverse values, add a data-quality check that flags columns with fewer than 1% unique values.
+| Field | Issue Type | Severity |
|---|---|---|
| mostly_null | +Missing Values | +Warning | +
| notes | +Missing Values | +Warning | +
| revenue | +Suspected Duplicate IDs | +Warning | +
| constant_col | +Near-Constant Column | +Info | +
| discount_pct | +Missing Values | +Info | +
| is_renewal | +Near-Constant Column | +Info | +
| order_status | +Near-Constant Column | +Info | +