Data & Analysis

How Datasets Are Built: From Raw to Ready

Follow data through every stage of the pipeline, from messy raw sources to clean, structured, analysis-ready datasets. Click each stage to explore and solve real challenges.

0 of 5 stages completed

Raw Source

Where data lives before anyone touches it

sales_FINAL_v2.csv

Date,Product,Rev???

2024-01,Widget??,$1,240

2024-01,Widget??,???

GET /api/users

{"users":[

{???name???:???Alice???

"id":null,"email":""}]}

form_log.txt

Name: Alice Sm??h

Email: alic@

Phone: ???-???-????

sensor_data.txt

10:23:01 temp=23.5

10:23:02 temp=ERR

10:23:03 --------

Raw data is data in its original, unprocessed form, pulled directly from the source with no cleaning, standardization, or transformation applied. It is almost never analysis-ready.

📋

Web Forms

User-submitted data with typos, missing fields, inconsistent formats.

🔌

APIs

JSON responses from third-party services; structures change without warning.

📡

Sensors / IoT

Time-series readings with gaps, sensor errors, and duplicate timestamps.

🗄️

Legacy Databases

Old tables with deprecated columns, mixed encodings, and undocumented fields.

Common problems in raw data

No standardizationDuplicate recordsMissing valuesMixed formatsEncoding issues

Your Turn

Click on cells that contain data quality problems. Find all 3 issues. 0/3 found

id	name	city	revenue	date
1	Alice	New York	4500	2024-03-15
2	Bob	London	null	2024-04-22
3	Carol	Tokyo	8200	12/01/2024
4	Dave	Berlin	3100	2024-06-10
2	Bob	London	null	2024-04-22
5	Eve	Paris	6700	2024-07-03

Collection

Pulling data from multiple sources into one place

Cleaning

Fixing errors, filling gaps, removing noise

Transformation

Reshaping and enriching data for analysis

Analysis-Ready

Clean, structured, and ready for insights