有無人可以簡單講咩係P-Value?

點解做Hypothesis testing, 當P-value少於significance level 就可以reject null hypothesis?

無上堂？

唔明喎

多謝喎

唔好理點解啦
佢話係就係
堆t test f test 數字0.425嗰啲都唔明嫁啦

用type I error 嘅方向諗會易明啲

唔想一知半解

做 Hypothesis test 就係想研究/搵到一d abnormal 情況 所以我地會想reject null hypothesis (normal case/facts)
假設significance level = 0.05, 如果p-value 少過0.05 啫係佢係一個abnormal situation, so reject null hypothesis

P-value細過significance level =出界 = reject H0

如果仲係degree就唔好去明啦
同你中學cos tan 兀嗰啲一樣
讀上去先去明啦
如果淨係做下metric, forecasting 用到就算真係

簡單嚟講
p value就係個probability of the null hypothesis is true

想reject個null
因為呢個世界有好多"就係咁岩得咁橋!"
咁你就set個可能性
例如
有95% 既可能性你得出呢個結果係咁岩得咁橋 唔係真係因為你加落去個effect而令到個結果變成咁 (alpha level=0.5)
p-value係個 cut off point
係過左呢個cut off 就係果95%以外
所以過左呢個p value
你就Conclude (其實正確黎講係assume 因為科學黎講你永遠唔知係咪真相)你既結果唔係"咁岩得咁橋"(即是reject個null)

有講錯請其他巴絲修正
無掂stat一排

要明best estimator同random sampling先解得通
因為所有test statistics 都牽涉到

標準textbook 答案

hi 啱啱做完project ??

P-value 就係 assume null hypothesis is true 既時候
你個sample觀察到extreme case的probability
而alpha 係我地人為set既rejection region，通常為0.05
假設依家alpha係0.05，H0 is true, 如果係one tail test 的情況下。
咁我地一次性抽sample 只係得5% 會落係rejection region
如果發生咗呢D extreme case，通常我地會覺得好小機會會發生呢種事情，但咁好彩都被我地撞中咗， 所以要reject H0

所以如果P < alpha ，即係話我地sample 觀察到個result比我地set既reject region 更加極端更加罕有，我地越有信心去reject H0。

利申： 畢業好耐，有錯輕屌！

唔該唔識唔好鳩噏
P value 係conditional probability 黎，講緊given null hypothesis is true嘅時候， 觀察到你手頭上或更extreme嘅data嘅probability，完全唔等如probability of null hypothesis being true。

sor9, probability of null distribution being true

有諗過, 係唔係因為當P-Value大於significance level, Type I error就會增加, 所以就唔可以reject null hypothesis