The guided cleaning dialog

The Step 2 tables are the fastest way to fix missing values and exact duplicates, but some data problems are subtler — dates hiding as text, look-alike categories, whitespace, statistical outliers. For those, Dotwave offers a guided cleaning dialog that detects issues automatically and walks you through them one at a time. This article explains how to open it, what it looks for, how the guided flow works, and how to skip anything you would rather leave alone.

How to open it

The guided dialog lives inside Step 2. Below the standard missing-values and duplicates tables you will find a link reading "Or use guided step-by-step cleaning". Clicking it opens the dialog, which runs its issue detector against your current data and presents whatever it finds. Because it reads the data as it stands at that moment, you can open it before you touch anything or after you have already resolved the obvious problems in the tables — it simply reports on the state the dataset is in when you launch it.

What the issue detector surfaces

When the dialog opens, its detector scans the dataset and raises the categories of problem that are easy to miss in a raw profile:

• Duplicates — exact repeated rows that should probably be collapsed.
• Missing values — columns that still contain nulls needing a fill or a drop.
• Dates stored as text — columns that look like dates but are typed as plain text, and so cannot be sorted or charted as dates until converted.
• Stray spaces — leading or trailing whitespace that makes otherwise-identical text values count as different entries.
• Look-alike categories — distinct labels that clearly mean the same thing and should be merged into one canonical value.
• IQR outliers — numeric values far enough outside the interquartile range to warrant capping or removal.

Each detected issue comes with the concrete fix Dotwave proposes, so you are never left to work out the remedy yourself — you are deciding whether to accept it.

How the one-issue-at-a-time flow works

Rather than dumping every problem on one screen, the dialog presents issues sequentially — one at a time. For the current issue it shows you what it found, which column is affected, and the fix it recommends. You review that single decision, apply the fix if you agree, and the dialog advances to the next issue. Working through problems one by one keeps each decision in focus and means you always understand exactly what a given click will change, instead of trying to reason about a dozen simultaneous edits. It also makes the resulting audit trail read as a clean, ordered sequence of deliberate steps.

How to skip an issue

Not every flagged issue is a problem you want fixed. Duplicate rows might be legitimate repeat orders; an apparent outlier might be a real, important value; a high-cardinality column might genuinely need all its distinct entries. When the dialog surfaces something you would rather leave as-is, you can skip it. Skipping moves you past that issue to the next one without applying any change and without removing anything from your data. The issue is simply left untouched, and because nothing was applied, nothing about that column is recorded as an edit. This lets you walk the full sequence of detected issues, accepting the fixes that make sense and passing over the ones that do not, and finish with only the changes you actually intended.