Cleaning studio

Build and simulate a repeatable data cleaning recipe. Preview results live and export the cleaned CSV in one click.



Transformation recipe

Select operations to apply to your dataset

Source dataset

STRUCTURAL CLEANUP

Remove completely empty columns Drops columns where 100% of rows are missing or spreadsheet export artifacts (e.g. Unnamed: 19) Remove duplicate records Preserves the first unique record and removes exact duplicate rows across all columns Remove constant / single-value columns Drops uninformative columns where all rows have the exact same value Normalize column names Converts headers to lowercase snake_case (e.g. "Body Type" → "body_type") and removes special characters

TEXT & STRING FORMATTING

Trim whitespace in text columns Strips accidental leading, trailing, and redundant spaces from text cells

Text Case Transformation:

OUTLIER & VALUE BOUNDING

Clip numeric outliers (1.5 × IQR boundary) Clamps extreme statistical anomalies to Q1 - 1.5*IQR and Q3 + 1.5*IQR without dropping rows

MISSING VALUE STRATEGY

Preserve missing values Leaves blank cells intact (recommended for initial diagnostic exploration) Impute missing values Fills numeric columns with their column median; fills text columns with "Unknown" Drop rows containing missing values Removes any row with one or more blank cells (may significantly reduce dataset volume)

PIPELINE

From messy to usable.

Import & Ingest Load raw CSV and perform UTF-8 structural validation

Transform & Clean Execute the configured cleaning and normalization recipe

Export & Verify Download clean UTF-8 CSV and verify with compare workbench



Data handling

All processing is executed request-locally in memory. No files or cleaning recipes are persisted to permanent cloud storage at this stage. Azure Blob Storage integration is planned for subsequent releases.

Cleaning studio

Transformation recipe

Recipe Results

Sample transformed records:

From messy to usable.

Data handling