| English | 释义 | 拼音 |
|---|---|---|
| conditional probability/kənˈdɪʃənl ˌprɒbəˈbɪlɪti/ | 条件概率 | tiáo jiàn gài lǜ |
A positive test
- In 1982 a famous study asked doctors: a disease affects 1% of people, a test catches 80% of cases but also flags about 10% of healthy people. If a patient tests positive, how likely is the disease? Most doctors said about 75%. The real answer was under 8%.
- Conditional probability 条件概率 depends on which group you are inside.
Restrict to the given group
- Class of 30: 18 cycle, 12 swim, 7 both. $P(\text{swim given cycle}) = \dfrac{7}{18}$: only the 18 cyclists count.
- $P(\text{cycle given swim}) = \dfrac{7}{12}$: now the 12 swimmers are the group.
In the class of 30 (18 cycle, 12 swim, 7 both), find P(cycles given swims).
难度 1/5 · 今日首次完成可获 100 金币及少量额外奖励
The group is the 12 swimmers.
Reversing the condition
- $\dfrac{7}{18} \ne \dfrac{7}{12}$: the numerator is the same, but the group (the denominator) changes.
- The whole-class $P(\text{swim}) = \dfrac{12}{30} = 0.4$ is a third, different number.
Conditional probability with tables and expected frequencies
Conditioning restricts the denominator to the given group
Classify the worked-case statements, then explain the units or invariant that justifies each decision.
Worked example: 1000 people
20% of people have a condition. A test is positive for 90% of people with it and 10% of people without it. A person tests positive. Find P(has the condition).
- Expected counts: of 1000, 200 have it → 180 positive; 800 do not → 80 positive.
- Restrict to positives: $180 + 80 = 260$.
- Answer: $\dfrac{180}{260} \approx 0.69$. Not 0.9: that is P(positive given condition).
已知条件改变分母所对应的群体
$P(A\mid B)=n(A\cap B)/n(B),\quad n(B)>0$
用本课的表格比较条件概率、反向条件概率和全体概率。高亮群体对应分母。
10% have a condition. A test is positive for 80% with it and 5% without it. Out of 1000 people, how many test positive?
难度 2/5 · 今日首次完成可获 140 金币及少量额外奖励
80 + 45.
For the last question, find P(has the condition given positive) to 2 d.p.
难度 2/5 · 今日首次完成可获 140 金币及少量额外奖励
80 ÷ 125.
Expected frequencies
- Turning probabilities into counts of 1000 or 10 000 people makes conditional questions clear.
- Fill a two-way table: condition / no condition × positive / negative, then read the column you need.
Of 200 students, 120 have a phone; 90 of those use it at night. Of the 80 without a phone, none do. Find P(has a phone given uses one at night).
难度 1/5 · 今日首次完成可获 100 金币及少量额外奖励
All 90 night users have phones.
The 1982 question
- 1% of 1000 = 10 sick, 8 test positive; 990 healthy, 99 test positive.
- $P(\text{sick given positive}) = \dfrac{8}{107} \approx 7.5\%$.
A disease affects 1%; the test catches 80% of cases and flags 10% of healthy people. Find P(sick given positive) as a percentage to 1 d.p.
难度 2/5 · 今日首次完成可获 140 金币及少量额外奖励
8 ÷ 107.
$P(\text{positive given condition})$ is not $P(\text{condition given positive})$. A test that is 90% accurate on sick people can still be wrong most of the time when it says positive, because healthy people far outnumber sick ones.
If a test is positive for 90% of people with a condition, then 90% of people who test positive have it.
难度 1/5 · 今日首次完成可获 100 金币及少量额外奖励
The groups are different.
For a conditional probability, restrict to the given group and use its total as the denominator. Turn probabilities into expected counts out of 1000, build the table, and read the right row or column. Reversing a condition changes the answer.
In a school, 60% of students take the bus. 25% of bus users are late at least once a term, and 10% of other students are. A late student is chosen. Find the probability they take the bus.
Conditional from counts.