Out-of-sample: woh data jisne rule banane mein hissa nahi liya.
Doosre naam: out of sample, out-of-sample, forward test, walk forward, holdout, in sample vs out of sample.
Ise kaise padhein
Rule banate waqt bahut se faisle hote hain: kaunsa threshold, kaunsi avadhi, kaunsa variation rakha jaaye. Har faisla us data ko dekh kar liya jaata hai jo saamne hai.
Isliye wahi data us rule ko jaanchne ke liye bekaar hai — usme jawab pehle se ghula hua hai. Out-of-sample data woh hissa hai jisne in mein se kisi faisle mein hissa nahi liya.
Yeh kya nahi batata
Ek out-of-sample khand khud ek sample hai, aur aksar chhota. Isliye kharab natija poora saboot nahi hai — jaise andar ka acha natija saboot nahi tha.
Bada khatra tareeke mein hai: agar out-of-sample dekh kar rule badla gaya, toh woh hissa ab andar ka hissa ban chuka hai. Uske baad wahi test dobara nahi kiya ja sakta.
Jude hue concepts
Bahut baar dekhne se kuch na kuch mil hi jaata hai.
Bina kisi signal wale data par bhi, bahut se rules aazmane par unme se sabse achha achha hi dikhega — kyunki aapne bahut baar dekha.
Lesson kholo