CSV Analyzer & Chart Generator

CSV and TSV — plain comma or tab-separated text files, the most common export format from Excel, Google Sheets, and virtually every database or system. XLSX/XLS support may come later; for now, export your spreadsheet as CSV first.

Yes — up to 100,000 rows, processed entirely in your browser. Larger files aren't supported, since going further would mean uploading your data to a server, which conflicts with keeping everything private.

No — parsing, statistics, charts, and everything else run entirely in your browser using JavaScript. Nothing about your file is ever sent to, or stored on, a server.

Count, missing, unique, min, max, sum, mean, median, standard deviation, and the 25th/75th percentiles — the same core set most spreadsheet or statistics tools report.

It's a 0-100 score weighted toward the issues that matter most: missing values and duplicate rows count for more than mixed data types or empty columns, since a missing or duplicated value silently skews every downstream statistic.

A row where every single value exactly matches another row in the file — not just a similar row, an identical one.

A column that's mostly one type (mostly numbers, say) but has a handful of values that don't fit — like a few "N/A" or "unknown" entries mixed into an otherwise numeric column. Those are flagged so they don't quietly skew calculations.

Each column's values are sampled and checked against number, date, and boolean patterns — if 90% or more of the non-empty values match one type, that's what the column is treated as. Anything short of that falls back to text.

Recommended charts are generated automatically from what's actually in your data — strong correlations, categorical breakdowns, group averages. The builder lets you pick exactly what you want: chart type, X and Y axes, aggregation, and an optional split-by column.

Sum, average, median, min, max, and count — pick whichever fits what you're trying to see, like average marks per department or total sales per month.

Yes — for bar, line, and area charts, select more than one Y-axis column (like Maths, Physics, and Chemistry together) to get one series per column, clustered or layered by group.

It adds a second categorical dimension, turning a plain bar chart into a grouped or stacked one — for example, average marks by department, split by gender, instead of department alone.

Every pair of numeric columns gets a Pearson correlation coefficient calculated; pairs with a coefficient of 0.5 or higher (a widely used threshold for a moderate-or-stronger relationship) are surfaced as scatter-plot recommendations and insight sentences.

No — they're templated directly from the same numbers already calculated for the stats, quality score, and correlations. That keeps them instant, free, and consistent, with nothing about your data sent to an AI model.

Yes — every chart, recommended or custom, has a download button that saves it as a PNG.

Yes — the Columns tab lists every statistic for every column at once, expandable to show all unique values for categorical columns.

Not for deep analysis or editing — it's built for a fast first look at a file: understanding what's in it, spotting problems, and getting a chart out, without opening spreadsheet software or writing a script.

Every file needs a header row as its first line — blank header cells get a placeholder name like "Column 3" so the rest of the file still parses correctly.