Help & user guide

If this is your first time, the flow below will guide you. Every operation is first previewed, shows what changes, and is only applied after you confirm; all of them are undoable.

1. Quick start

1. Open the workspace and upload a file (or pick a sample dataset).
2. In the import dialog check the parsing preview; change the encoding, delimiter or sheet if needed.
3. On the Profile & quality tab see the issues; the “Fix” buttons open the right operation with parameters pre-filled.
4. On the Clean and Transform tabs preview and apply operations.
5. Draw conclusions with Analysis, Chart and Validate; download with Export.

2. Import

• Formats: CSV, TSV, TXT, XLSX/XLSM, XLS, ODS, JSON, JSON Lines, XML, HTML tables, Parquet, Feather, Stata (.dta) and fixed-width text (.fwf). For multi-sheet Excel files you can pick one sheet or all sheets (each becomes its own dataset).
• Encoding and delimiter are detected automatically. If Turkish characters look broken, switch the encoding to “Windows-1254”.
• Your data is not corrupted: leading zeros such as “00123”, numbers longer than 19 digits and ambiguous dates are preserved as text. You convert numbers/dates yourself, with a preview.
• Cells containing only spaces and placeholders such as NULL, NaN, #N/A are treated as empty (you can turn this off). “NA” is deliberately left alone: it may be the country code for Namibia.

3. Profile and data quality

The quality score is built from completeness, uniqueness and consistency, and its formula is explained in the app. It is a guide, not an exact measure. For each issue the number of affected cells is shown.

4. Cleaning and transformation

• Preview → verify → apply: the preview shows the change in row/column counts, the number of changed cells, and before/after examples per column.
• Save as a new dataset: tick this for structure-changing operations (summaries, pivots, joins …) to keep the original.
• Undo / redo: use the buttons at the top or return to any step from the History tab.

Formulas (calculated columns)

Write columns in square brackets. Examples:

[Unit Price] * [Qty]
[Total] * 1.2
if([Score] >= 50, "Pass", "Fail")
upper(left([Name], 1)) + lower(mid([Name], 2, 50))
days_between([Order Date], [Delivery Date])
coalesce([Mobile], [Phone], "none")

The full list of functions is in the “Formula syntax and function list” section of Transform → Add calculated column. Formulas run in a safe interpreter; arbitrary code cannot be executed.

5. Analysis and statistics

Descriptive statistics, frequency/cross tables, correlation (Pearson, Spearman, Kendall) and tests (Welch t-test, Mann–Whitney, ANOVA, Kruskal–Wallis, chi-square, normality) are real calculations. Results come with plain-language interpretation and assumption warnings; statistical significance does not mean importance or causation.

6. Charts

As you choose fields the chart is aggregated on the server and drawn in your browser. For large data a scatter plot shows a random sample and says so explicitly. Download as SVG or PNG.

7. Validation

Rules for e-mail, URL, Turkish phone number, Turkish ID number (with the checksum algorithm), IBAN (mod 97), regex, numeric range, allowed values, length, number/date type and uniqueness. Invalid rows can be opened as a new dataset, flagged or deleted.

8. Comparison

See column differences, added/removed rows and changed cells between two datasets. If you pick a key column (e.g. ID), records are matched and every value that changed is listed at cell level.

9. SQL

Every dataset in your workspace can be queried as a table (SQLite syntax). Only SELECT/WITH works; writes, ATTACH, PRAGMA and extension loading are blocked, and there are time and row limits. Dates appear as ISO text.

10. Recipes (repeatable work)

From the History & recipe tab, download the operations you applied as a JSON recipe. Upload it to another file with the same structure (e.g. next month's report) and re-apply it in one click; it stops at the first error and reports.

11. Export

CSV, TSV, Excel, JSON, JSON Lines, Markdown, HTML and SQL. For Excel in Turkish locales choose the UTF-8 (BOM) encoding and a semicolon delimiter if needed. With “formula protection” on, texts starting with =, +, - or @ get a leading '.

Frequently asked questions

How long is my data kept?

You choose the retention period (1 hour – 7 days; it counts from your last activity and can be changed later) and the data is deleted automatically when it ends; you can delete everything at any time from the side panel. See the privacy page for details.

Do I need an account?

No. Your workspace is tied to a functional cookie in your browser. If you clear the cookie or use another browser you will not be able to reach the same data.

My file is too large.

If it exceeds the values on the limits page, split the file or export only the columns you need.

Can I trust the results?

Calculations are done with standard libraries (pandas, NumPy, SciPy) and every operation shows what changed. Still, before critical decisions we recommend cross-checking results with a method you know.