Hypothesis Testing Flashcards
1
Q
- Label purpose
A
- Signpost to tell us which data is distinct
2
Q
- Binary vs categorical vs continuous labels, give examples
A
- Binary – 0s and 1s (Yea/Nah)
- Categorical – distinct labels of types (colours)
- Continuous – range of values (temperature)
3
Q
- Consequence of bad or incorrect labelling
A
Misleading, biased outputs, ethical issues
4
Q
- If a particular label occurs rarely, how does that affect what can be learned or inferred?
A
Rare labels can lead to class imbalance, reduced model performance, overfitting, limited insights, and training challenges.
5
Q
- Purpose of hypothesis test
A
- Helps us make decisions about how we should work with data
6
Q
- How hypothesis testing be useful in a brute force approach assessing patterns in data?
A
- Hypothesis testing helps us determine which pattern or feature is best.
7
Q
- Risks of running many hypothesis tests?
A
- P-hacking
8
Q
- Two types of errors and how they inhibit data analysis?
A
-Type 1 - mistaken significance
-Type 2 - Missed significance
