Review: Data Skills

You've been through a lot of material. Before the capstone, let's lock it in. Testing yourself beats rereading, and that's been shown again and again. So pause on each question, and commit to an answer before you hear it.

• Lock it in
• Testing beats rereading
• Commit to each answer

Recap: wrangling

Wide tables are for reading, and long tables are for computing. Use category types, chunks, and parquet for big files. Always ask why a value's missing, as dropping rows can bias your results. And investigate outliers before deleting them, as some of them are your best customers.

Question 1

Question 1. Which format suits grouping and plotting? A, Wide. B, Long. C, Merged cells. D, PDF. Pause and choose. The answer is B. One row for each measurement.

Question 2

Question 2. Why is parquet better than CSV for analysis? A, It is text. B, It stores columns, compressed, with types. C, It is older. D, It needs no code. Pause and choose. The answer is B. You can read just the columns that you need.

Question 3

Question 3. High earners skip the income question. What happens if you drop those rows? A, Nothing. B, Your averages become biased. C, Files get smaller, that is all. D, It fixes the data. Pause and choose. The answer is B. Ours fell from twelve to eleven.

Question 4

Question 4. Which statistic resists outliers? A, Mean. B, Median. C, Maximum. D, Sum. Pause and choose. The answer is B. Ours barely moved.

Question 5

Question 5. A rolling seven day mean removes what? A, Weekly rhythm. B, All trends. C, Outliers fully. D, Holidays. Pause and choose. The answer is A. Each window holds one of each day.

Question 6

Question 6. Why are funnel steps worth the same when lifted by ten percent? A, They multiply. B, They add. C, They are random. D, They are equal. Pause and choose. The answer is A. So improve the easiest step.

Summary