https://www.youtube.com/watch?v=yNt3NnbsIzQ&list=PLsri7w6p16vtEz_J1G7HQG-Rm8vpZPtPS&index=2&t=0s ================================================================================ * Hypothesis * Claim or guess about the "given fact" or "fact that you want to prove" * Given fact: drink A has 300ml * "fact that you want to prove": drink A has really 300ml? ================================================================================ * There are guesses which are correct and incorrect * so you have 2 hypothesises * null hypothesis * anti hypothesis ================================================================================ * research: estimate parameters of population via sample * opinion on parameters: hypothesis ================================================================================ null hypothesis = zero hypothesis null hypothesis: $$$H_0$$$ null hypothesis is the hypothesis that you don't need to research sentence a: "drink A has 300ml" if sentence a is correct, you don't need to perform research on it ================================================================================ anti-hypothesis: is hypothesis against "null hypothesis" anti-hypothesis is used for research anti-hypothesis: $$$H_1$$$ sentence b: "drink A doesn't have 300ml" ================================================================================ Test on hypothesis set p-value $$$\alpha$$$ value from hypothesis $$$\subset$$$ $$$\alpha$$$: hypothesis is selected value from hypothesis $$$\notin$$$ $$$\alpha$$$: hypothesis is rejected $$$\alpha$$$, value from hypothesis, $$$\alpha$$$: two tail test value from hypothesis, $$$\alpha$$$: one tail test ================================================================================ * Test "anti hypothesis" which claims "null hypothesis" is not true * If "anti hypothesis" is correctly proved, "null hypothesis" is rejected and "anti hypothesis" is selected * Meaning: if drink A has 300.1ml, 310ml, etc, you accept "null hypothesis" (drink A has 300ml) If drink A has 200ml or 400ml, you reject "null hypothesis" ================================================================================ * Generally, customer doesn't have complain if drink A has more than 300ml * So, in that case, you can perform "one-sided test" * In case of drink A example, you can perform only left-one-sided test ================================================================================ Workflow for test 1. Set hypothesises $$$H_0$$$, $$$H_1$$$ 2. Set p-value 3. Set rejection area 4. Calculate statistical value 5. Decision