Null model: kuch na hone par output kaisa dikhta hai.
Doosre naam: null model, shuffle test, permutation test, randomisation test, null hypothesis, the null.
Ise kaise padhein
Kisi rule ka natija akela dekh kar yeh nahi pata chalta ki woh acha hai ya nahi. Sawaal hamesha yeh hai: itna acha toh yun hi kitni baar aa jaata?
Null model wahi jawab deta hai. Wahi data, wahi tareeka, lekin rule ka asar jaan bujh kar hataya hua. Kai baar chalane par ek daayra banta hai, aur asli natije ko us daayre ke saamne rakha jaata hai.
Yeh kya nahi batata
Null model utna hi acha hai jitna uska shuffle todta hai. Agar shuffle samay ka kram bacha deta hai, toh underlying ka apna badhna bacha reh jaata hai.
Us haalat mein ek rule jo bas badhte bazaar ke saath khada tha, woh null model ko bhi haraa sakta hai — aur natija asli lagne lagta hai jabki woh sirf bazaar hai.
Jude hue concepts
Shuffle test: natije ko kis ke saamne rakha jaaye.
Akela number pe faisla nahi ho sakta. Shuffle test wahi cheez banata hai jiske saamne us number ko rakha ja sake.
Lesson kholo