Foundations — Expected value, variance, standard deviation — properties

4.9.4 · D1 · Maths › Probability Theory & Statistics › Expected value, variance, standard deviation — properties

Isse pehle ki tum Expected value, variance, standard deviation — properties mein koi bhi formula padh sako, tumhe har ek symbol ko zero se samajhna hoga jo woh tumpe throw karta hai. Neeche, har idea kuch nahi se build kiya gaya hai, ek picture se joda gaya hai, aur justify kiya gaya hai — phir usse agla idea khada hota hai.

1. Ek number line aur uske upar ek point

Sab kuch number line se shuru hota hai: ek horizontal ruler jahan har position ek number hai. Chhote numbers left pe, bade numbers right pe, beech mein $0$.

Us line par ek point ek specific outcome hai — jaise, "die ne $4$ dikhaya," jise hum position $4$ pe mark karte hain.

Humein yeh isliye chahiye kyunki ek random variable ka kaam is ruler pe kahin pe land karna hai, aur humari dono summaries (center, spread) us par locations aur distances ki tarah describe ki jaati hain.

2. Symbol $X$ — ek random variable

Picture: socho ek dart jo ruler pe kahin land karega. $X$ hai "jahan bhi dart land kare." Die roll karo → $X$ woh face hai. Coin flip karo aur heads ko $1$ score karo → $X$ ya $0$ hai ya $1$.

Topic ko yeh kyun chahiye: saare formulas — $\mathbb{E}[X]$, $\mathrm{Var}(X)$ — is dart ki aadat ke baare mein sawal hain. Capital letters jaise $X,Y$ ka matlab hamesha "ek random number" hota hai. Lowercase $x_i$ ka matlab hai "ek specific possible value jo yeh le sakta hai."

3. Probability $p_i$ — har value pe kitna mass baitha hai

Picture: har possible value pe clay ka ek lump rakho. Uncha lump = zyada likely. Total clay ka wazan exactly $1$ hona chahiye — kyunki kuch na kuch hamesha hota hai.

Symbol ${\sum }_i$ (capital Greek sigma) ka matlab bas "saari possibilities $i$ pe add up karo" hai. ${\sum }_i{p}_i$ padho jaise "har lump ka wazan jodo." Woh total $1$ hai — yeh fact parent note mein terms ko collapse karne ke liye baar baar use hota hai.

Continuous variables ke liye (jo koi bhi real number ho sakti hain, sirf ek list nahi), clay ek smooth curve $f(x)$ mein smear hoti hai jise density kehte hain. Tab "add up karna" ek integral $\int$ ban jaata hai, jo bas infinitely many infinitely-thin slices pe ek sum hai.

4. Expected value $\mathbb{E}[X]$ aur uska symbol $\mu$

Picture: ruler ke neeche apni ungali rakho aur slide karo jab tak koi bhi side tip na kare. Woh ungali ki position $\mathbb{E}[X]=\mu$ hai.

"Opposite × mass" kyun? Har lump balance ko apni position $x_i$ ki taraf kheenchta hai, apne weight $p_i$ ke barabar kheenchne ki strength ke saath. Mass se weighted position ka average ${\sum }_i{x}_i{p}_i$ deta hai — physics ka center of mass. Isliye hi parent mein formula waise dikhta hai.

5. Mean se distance, aur hum kyun square karte hain

Spread ki baat karne ke liye, hum poochte hain: ek typical outcome center $\mu$ se kitni door hai?

Kisi value ka deviation hai $x_i-\mu$: yeh balance point se kitna right ($+$) ya left ($-$) baitha hai.

Hum signs square karke hataate hain: $(x_i-\mu {)}^2$ hamesha $\ge 0$ hota hai, aur yeh door waale points ko paas waale points se zyada punish karta hai ($3$ ki deviation $9$ contribute karti hai).

6. Variance $\mathrm{Var}(X)$ aur standard deviation $\sigma$

Picture: har lump ke liye, ungali tak ki distance measure karo, use square karo, phir un squares ka mass-weighted average lo. Wide spread → bade squared distances → bada variance.

Units problem: agar $X$ metres mein hai, toh squared distances m² mein hain. Wapas plain metres pe aane ke liye hum square root lete hain.

Symbol $\sigma$ (Greek "sigma", lowercase) hamesha standard deviation mean karta hai; isliye ${\sigma }^2$ ka matlab variance hai. Yahi reason hai ki parent dono freely likhta hai.

7. $X$ ke functions, aur symbol $g(X)$

Kabhi kabhi hum $X$ ki nahi balki ek rule ki parwah karte hain jo usse apply ho — us rule ko $g$ kaho. Toh $g(X)$ hai "jahan bhi $X$ land kare, usse transform karo."

• $g(x)=x^2$ → $g(X)=X^2$ (variance ke liye use hota hai).
• $g(x)=ax+b$ → ek straight-line rescale (scale/shift rules ke liye use hota hai).

Topic ko yeh kyun chahiye: computational formula mein $\mathbb{E}[X^2]$ exactly LOTUS hai $g(x)=x^2$ ke saath. Is shortcut ke bina tumhe $X^2$ ka poora distribution rebuild karna padta.

8. Ek doosra variable $Y$, aur do variables kaise relate karte hain

Randomness add karne ke liye (jaise coin flips ka sum), humein ek doosra random variable $Y$ chahiye: usi ruler pe ek aur dart.

Poori detail Covariance and Correlation mein hai; yahan tumhe bas symbol pehchanna hai aur jaanna hai ki independence ise vanish kar deti hai — isliye sum-of-variances rule ko independence chahiye.

9. Prerequisite map

Ise top-down padho: geometry (number line) random variable aur probability mass ko feed karta hai; woh mean ko feed karte hain; mean deviations, squares, variance, aur finally standard deviation ko feed karta hai. Functions aur ek doosra variable LOTUS aur sum rule ko power karne ke liye branch off karte hain. Yeh foundations Law of Large Numbers aur Central Limit Theorem se aage connect hote hain, jo describe karte hain ki jab tum bahut saare samples collect karo toh means aur spreads ka kya hota hai.

Equipment checklist

Right side cover karo aur apne aap ko test karo.