For students learning data science
From raw data to a dataset you can use.
DataCatalyst helps you inspect datasets, identify quality issues, and prepare a clean copy through a guided, step-by-step workflow.
- Rule-based checks, with the evidence shown
- You approve every change
- Your original file stays unchanged
How it works
What happens to your file
Findings come with the evidence behind them, such as the rows and values involved. Nothing is changed until you preview a change and approve it.
Upload
Choose or drag in a CSV file. It is checked and previewed before anything is analyzed.
See the findings
See what needs attention: missing values, duplicate rows, mismatched types, date formats, stray spaces, out-of-range or unusual numbers, and category labels written in different ways.
Review suggestions
Each suggestion shows what was found, the evidence behind it, and what to be careful about.
Preview changes
Work on one issue at a time and see every value that would change before you approve it.
Download
Save a prepared copy as a new CSV file, plus a JSON report of the findings.
Features
What it checks and prepares
Everything listed here works in the current version and runs on the file you upload.
-
CSV upload and preview
Drag and drop or choose a file up to 10 MiB, then look at its first rows before the analysis starts.
-
Missing values
Blank cells, absent fields and markers such as "NA" or "None", counted for every column.
-
Duplicate rows
Rows that exactly repeat an earlier row, each listed with the row it copies.
-
Type inconsistencies
Values that do not match a column's likely type, such as text in a column of numbers.
-
Date formats and spacing
Dates written in more than one format, and values with stray spaces at the start or end.
-
Ranges and possible outliers
Values outside a known range, like a percentage above 100, and unusually high or low numbers worth a second look.
-
Category labels
Labels that may be one category written in different ways, such as "Computer science" and "Computer Science", or "Freshman" and "First-year". They are flagged for you to review, never merged.
-
Rule-based recommendations
Prioritized suggestions with their evidence and cautions, plus optional guidance for your experience level and goals.
-
Guided preparation
Fill in or remove missing values, drop chosen duplicate rows, trim spaces, standardize dates, and convert numbers written as words. Compare values before and after, approve the change, then download the prepared CSV.
About DataCatalyst
What to expect from DataCatalyst
DataCatalyst is an early-stage learning tool for people starting out in data science, machine learning and data analysis. It points out what to check and shows the evidence for each finding; the decisions about your data stay with you.
-
Rule-based checks
Every finding and recommendation comes from fixed, documented rules. You can open the evidence and the rule behind each one.
-
Suggestions, not corrections
Recommendations point to what is worth reviewing. They are not guaranteed fixes, and a flagged value may be correct.
-
You review and approve
Nothing changes until you preview a change and approve it. Only then can you download the prepared copy.
-
Your original file is never changed
Changes go into a separate copy. The file you uploaded is never changed. Your file is not permanently stored by DataCatalyst. The server processes it for each request. Files larger than 1 MiB may be held briefly in a temporary file while a request is being handled.
Get started
Upload a CSV file
You will see its first rows and can choose how explanations are shown before anything is analyzed.
You have a file open in your workspace. Uploading a file here replaces it. To go back to it instead, use Return to your analysis at the top of the page.