AIF-C01 ยท AI and ML basics
Labeled vs unlabeled, structured vs unstructured
Labeled versus unlabeled, structured versus unstructured, tabular, time-series, image and text data and which fit which task.
Key points
- Labeled data: each example has the correct answer (an email marked spam); needed for supervised learning.
- Unlabeled data: no answer attached; suits pattern discovery.
- Structured data: rows and columns, such as tabular records and time-series with timestamps.
- Unstructured data: no fixed schema, such as free text, images, audio, video.
Practice questions
What makes a dataset 'labeled'?
- Each example is paired with the correct output
- It is stored in a CSV file
- It contains more than a million rows
- It has been encrypted
Show answer
Answer: A. Each example is paired with the correct output
Labeled means each example carries the target answer. Size, format or storage location do not matter.
Which of these are structured data?
- A relational table of customer orders
- Voice recordings of support calls
- A CSV file with columns for date, store and revenue
- Scanned photos of receipts
- Free-text product reviews
Show answer
Answer: A. A relational table of customer orders, C. A CSV file with columns for date, store and revenue
Rows in a relational table and a CSV of order records both have a fixed schema. Audio and scanned photos are unstructured; a free-text review is unstructured as well.
True or false: unlabeled data is useless for machine learning.
Show answer
Answer: False
False. Unlabeled data is used for clustering, anomaly detection and learning representations, and large models are pre-trained on it.
Practice 8 questions on this topic in Bitzu
Short lessons, instant explanations and spaced review of what you get wrong. Free on iOS and Android.
