You have a column called "country" with 180,000 records. You assume it's mostly "United States" with some international records. A distribution analysis shows the reality: 45% are "United States", 12% are "US", 8% are "U.S.", 6% are "USA", and the remaining 29% are spread across 40 other values. You don't have a clean country column — you have a standardization disaster. That's what distribution analysis reveals.
Value distribution analysis examines how values are spread across a column — what the most common values are, how many distinct values exist, and whether the distribution looks like what you'd expect given the column's purpose.
What Distribution Analysis Reveals
Standardization problems — When a column that should have a small, controlled set of values (like country, status, or category) has dozens or hundreds of variants, you have a standardization failure.
Data entry patterns — Spikes at specific values (like "N/A", "Unknown", or "Other") reveal how users handle missing or uncertain information — often hiding the true extent of missing data.
Outlier clusters — Values that appear many times but shouldn't — like a revenue column where "0" appears 3,000 times in a dataset of 10,000 records — signal either a business condition or a data quality problem.
Sohovi profiles your datasets for quality issues in minutes — see what's broken before it breaks your pipeline — try Sohovi free.
Format inconsistencies — A date column with 12 distinct date formats reveals that multiple systems or entry methods contributed to the dataset.
Skewed distributions in numeric columns — If 90% of your numeric values cluster around 10–20 but 10% are in the millions, you may have a unit mismatch error.
How to Read a Distribution
For categorical columns (limited set of values): a clean distribution shows a small number of distinct values each appearing many times. A problematic distribution shows many distinct values that are likely variations of the same thing.
For numeric columns: a histogram reveals whether values follow an expected pattern (bell curve, uniform, right-skewed) and shows where outliers cluster.
For text columns with high cardinality (many distinct values, like names or descriptions): distribution analysis is less useful for individual values, but looking at value length distribution and pattern matching reveals format issues.
Sohovi displays the top values and distinct value counts for every column in your uploaded CSV — making distribution analysis instant and visual, without needing to write pivot tables or formulas.
Using Distribution Analysis Practically
When you see an unexpected distribution, ask: "Does this reflect reality, or does it reflect a data quality problem?" A country column with 47 variants almost certainly reflects a standardization problem, not 47 genuinely different country formats in your customer base.
Distribution analysis is the fastest way to find hidden data quality problems that don't show up as missing values or format errors.
Sohovi profiles every column in your dataset for completeness and flags the exact rows where values are missing — free to try.
