- Hypothesis testing:

    - Layman term : Method of testing if an assumption falls in feasible likelihood estimation or just a non-sense.

    - Hypothesis : statement about an unknown population parameter (Your guess of the average value of a sample)

    - Hypothesis test : test of two competing hypotheses. 

        - Null hypothesis : Your guess

        - Alternative hypothesis : Challenging idea against your guess

    - 3 tests: 

        - direction of test = direction of alternative hypothesis (or challenging idea)

        - hypothesis for less or fewer = left tailed test

        - hypothesis for greater or more = right tailed test

        - hypothesis for different = two tailed test

    - Testing Method : Compute sample mean and see the difference between guessed mean and sample mean. Measure if the difference is significant or not.

    - Solution:

        - z-test

        - proportion z-test

        - t-test

        - ANOVA

        - chi-square test of independence

        - Chi-square goodness of fit

        - Non-parametric Wilcoxon-signed rank test 

        - Non-parametric Wilcoxon-Mann-Whitney test

        - Non-parametric Kruskal-Wallis test

- A/B Testing : Hypothesis testing on 2 different scenarios (control vs treatment group)

- p-value : 

    - probability of obtaining a result if the null hypothesis is true

    - probability value from z-score or significance level

    - Large p-value = p values not on tails = Large support for null hypothesis (p_value > alpha)

    - Small p-value = p values on tails = No support for null hypothesis (p_value <= alpha)

    - cut-off point = common significance level is 5%

    - confidence interval = 1- alpha = 95%

    - if the hypothesized population parameter is within the confidence interval, you should fail to reject the null hypothesis.

- Null is False / Negative.

- False Positive = Type 1 error

- False Negative = Type 2 error