How a Column's Type Gets Detected From Its Data
There's no header metadata declaring "this column is numeric" in a plain CSV file - the type has to be inferred from the values actually present. This tool checks every non-blank value in a column and treats it as numeric if at least 60% of them parse cleanly as numbers, deliberately NOT requiring 100% - a handful of "N/A" or "unknown" values scattered through an otherwise numeric Salary column shouldn't flip the whole column to "text" and lose the ability to flag those specific entries as mismatches.
That 60% threshold is a practical compromise: strict enough to correctly classify a genuinely mixed-content column as text, loose enough to tolerate the occasional real-world data-entry exception in an otherwise clean numeric column.
Why a Suggested Fill Is a Statistical Guess, Not a Fact
Filling a blank numeric cell with the column's average, or a blank text cell with its most common value, is a reasonable STATISTICAL default when you genuinely don't know the real answer - but it's never actually true for that specific missing record. The real value for a missing Age cell was some specific number, not necessarily the dataset's average age.
That's why this tool always shows suggested fills as a visible, reviewable overlay rather than silently rewriting your data - the goal is to make gaps and their best-guess fill obvious at a glance, not to auto-correct your dataset without your knowledge.
Frequently Asked Questions
There's no safe way to guess what a mismatched value SHOULD have been (a stray "unknown" in a Salary column could mean many different things), so the tool limits itself to reliably flagging the problem for a human to resolve, rather than making an unreviewed change that could introduce a worse error.
Not necessarily - for a skewed distribution (a few very large or very small values pulling the average away from most of the data), the median can be a more representative fill than the mean. This tool uses the average specifically because it's the more common, simpler default; always sanity-check whether it makes sense for your particular column.
A blank cell is genuinely empty - nothing was recorded. A type mismatch is a cell that DOES have a value, but that value doesn't match the pattern of the rest of its column (like text inside an otherwise numeric column) - the two are flagged with different visual highlighting since they represent different underlying data problems.
Be especially cautious there - a statistically-plausible fill is fundamentally an estimate standing in for missing information, which is a reasonable stopgap for many uses but generally not appropriate to present as verified, accurate data in a context where precision genuinely matters.