All language subtitles for 01 - Introduce Bayesian analysis

- Previously in this course, I have discussed

what are known as descriptive statistics.

Descriptive statistics, as the name implies,

provide facts about your data.

Those can include measures

such as medians, means, variance and standard deviation.

You can then use these facts about your data

to make estimates at a known confidence level.

For example, you might think that most of your customers

live within a 25 mile radius,

plus or minus three or four miles.

And you might be wondering, what else is there?

It turns out there's a lot more. Let me give you an example.

Let's assume, for the sake of argument,

that you might have the flu.

And your test for the flu has come back positive.

The test has the following characteristics

and all of these will be important for our analysis.

It returns the correct result 85% of the time.

It identifies healthy individuals

as having the flu 10% of the time.

So in other words, if you do have the flu,

you'll get a positive result 85% of the time.

A negative result 15% of the time.

If you don't have the flu,

then you'll get a positive result 10% of the time,

a false positive.

And you'll get the correct result of negative the other 90%.

We also have a base rate which assumes

that about 1% of the population

actually has the flu at the time you take the test.

So the question is,

what is the probability that you actually have the flu?

For this, we use Bayes' Rule.

We combine the accuracy, false positive and base rate

to find the result.

So if 1% of the population has the flu,

and the test is 85% accurate,

but there's also a 10% false positive rate,

a positive test means the probability

that you actually have the flu is

7.91%.

How we got to that number

is the subject of the rest of this chapter.