Statistics - Day 8

Alright! So, today we will study about Hypothesis testing.

Hypothesis testing is like dividing the problem you want to test into whether it is true or is false.

H0 (null hypothesis) is the one you want to reject - opposite of what you want to test

H1 (Alternative hypothesis) anything other than the null hypothesis.

So, how do you start with the Hypothesis testing?

1. Calculate a statistic (x mean)

2. Scale it (CLT) and calculate z score (sample mean- pop mean/s/sqrt(n))

3. Check if z is in rejection region (usually more than 2 std dev away from mean)

There is also a one sided test when the hypotheses don't contain the equality sign.

Type 1,2 errors:

Type 1 error: When you reject true null hypothesis

Type 2 error: When you accept false null hypothesis

You should reject null hypothesis if:

absolute value of Z-score>positive critical value of z

But there also needs to be a measure using which we can still reject the null hypothesis - called "P-value"

It is the smallest level of significance at which we can reject null hypothesis.

More on this tomorrow!