Intuition The big picture (WHY these five?)Continuous distributions describe quantities that can take any value in a range. The five here are the "workhorses" because they each model a distinct physical story :
Uniform — total ignorance inside an interval ("every point equally likely").Normal — sums/averages of many small effects (Central Limit Theorem).Exponential — waiting time until the first random event (memoryless).Gamma — waiting time until the k-th event (sum of exponentials).Beta — a random probability / proportion living in [ 0 , 1 ] [0,1] [ 0 , 1 ]
Learn them as a family tree , not five isolated facts. That's the 80/20: master the stories and the rest is algebra.
A continuous random variable X X X probability density function f ( x ) ≥ 0 f(x)\ge 0 f ( x ) ≥ 0 ∫ − ∞ ∞ f ( x ) d x = 1 \int_{-\infty}^{\infty} f(x)\,dx = 1 ∫ − ∞ ∞ f ( x ) d x = 1 cumulative distribution function F ( x ) = P ( X ≤ x ) = ∫ − ∞ x f ( t ) d t F(x)=P(X\le x)=\int_{-\infty}^{x} f(t)\,dt F ( x ) = P ( X ≤ x ) = ∫ − ∞ x f ( t ) d t f ( x ) = F ′ ( x ) f(x)=F'(x) f ( x ) = F ′ ( x )
Definition The Gamma function (our integration tool)Γ ( α ) = ∫ 0 ∞ t α − 1 e − t d t , α > 0 \Gamma(\alpha)=\int_0^\infty t^{\alpha-1}e^{-t}\,dt,\qquad \alpha>0 Γ ( α ) = ∫ 0 ∞ t α − 1 e − t d t , α > 0 derive on the fly : Γ ( α + 1 ) = α Γ ( α ) \Gamma(\alpha+1)=\alpha\,\Gamma(\alpha) Γ ( α + 1 ) = α Γ ( α ) Γ ( 1 ) = 1 \Gamma(1)=1 Γ ( 1 ) = 1 Γ ( n ) = ( n − 1 ) ! \Gamma(n)=(n-1)! Γ ( n ) = ( n − 1 )! Γ ( 1 2 ) = π \Gamma(\tfrac12)=\sqrt\pi Γ ( 2 1 ) = π
U ( 2 , 8 ) U(2,8) U ( 2 , 8 ) μ = 2 + 8 2 = 5 \mu=\frac{2+8}{2}=5 μ = 2 2 + 8 = 5 Why? midpoint. Var = ( 8 − 2 ) 2 12 = 3 \operatorname{Var}=\frac{(8-2)^2}{12}=3 Var = 12 ( 8 − 2 ) 2 = 3 P ( X ≤ 5 ) = 5 − 2 8 − 2 = 1 2 P(X\le 5)=\frac{5-2}{8-2}=\tfrac12 P ( X ≤ 5 ) = 8 − 2 5 − 2 = 2 1 Why this step? CDF of uniform is linear: x − a b − a \frac{x-a}{b-a} b − a x − a
Time you wait for the first event in a Poisson process with rate λ \lambda λ memoryless : having waited 5 minutes already tells you nothing about the future wait.
Worked example Bus arrives at rateλ = 0.5 \lambda=0.5 λ = 0.5 = 1 / 0.5 = 2 =1/0.5=2 = 1/0.5 = 2 P ( wait > 3 ) = e − 0.5 ⋅ 3 = e − 1.5 ≈ 0.223 P(\text{wait}>3)=e^{-0.5\cdot 3}=e^{-1.5}\approx 0.223 P ( wait > 3 ) = e − 0.5 ⋅ 3 = e − 1.5 ≈ 0.223 Why this step? survival = e − λ t =e^{-\lambda t} = e − λ t
Wait for the α \alpha α α = k \alpha=k α = k X X X k k k Exp ( λ ) \text{Exp}(\lambda) Exp ( λ ) Exponential is Gamma with α = 1 \alpha=1 α = 1 .
Gamma ( 3 , 2 ) \text{Gamma}(3,2) Gamma ( 3 , 2 ) = 3 / 2 = 1.5 =3/2=1.5 = 3/2 = 1.5 = 3 / 4 = 0.75 =3/4=0.75 = 3/4 = 0.75 Why? sum of 3 exponentials each mean 1 / 2 1/2 1/2 3 × 1 2 = 1.5 3\times \tfrac12=1.5 3 × 2 1 = 1.5 3 × 1 4 3\times\tfrac14 3 × 4 1
The Central Limit Theorem : add up many independent small influences and the total is approximately bell-shaped, regardless of the originals. That's why measurement errors, heights, and exam scores look Normal.
N ( 170 , 10 2 ) N(170,10^2) N ( 170 , 1 0 2 ) P ( X ≤ 185 ) P(X\le 185) P ( X ≤ 185 ) z = 185 − 170 10 = 1.5 z=\frac{185-170}{10}=1.5 z = 10 185 − 170 = 1.5 Why this step? put on standard scale. Φ ( 1.5 ) ≈ 0.933 \Phi(1.5)\approx 0.933 Φ ( 1.5 ) ≈ 0.933 P ( ∣ Z ∣ < 1 ) ≈ 0.68 P(|Z|<1)\approx0.68 P ( ∣ Z ∣ < 1 ) ≈ 0.68
A random number between 0 and 1 — perfect for modeling an unknown probability or proportion . Shape is hugely flexible: α = β = 1 \alpha=\beta=1 α = β = 1 α , β > 1 \alpha,\beta>1 α , β > 1 α , β < 1 \alpha,\beta<1 α , β < 1
Beta ( 2 , 5 ) \text{Beta}(2,5) Beta ( 2 , 5 ) = 2 7 ≈ 0.286 =\frac{2}{7}\approx0.286 = 7 2 ≈ 0.286 β > α \beta>\alpha β > α Why? more "failure mass" pushes the average left.
Common mistake "Exponential rateλ \lambda λ Why it feels right: λ \lambda λ Fix: λ \lambda λ rate ; the mean is the reciprocal 1 / λ 1/\lambda 1/ λ
Common mistake "Variance ofU ( a , b ) U(a,b) U ( a , b ) ( b − a ) 2 / 4 (b-a)^2/4 ( b − a ) 2 /4 Why it feels right: the range is b − a b-a b − a Fix: the integral gives ( b − a ) 2 / 12 (b-a)^2/\mathbf{12} ( b − a ) 2 / 12 1 3 − 1 4 \frac13 - \frac14 3 1 − 4 1
Common mistake "I can readN ( μ , σ 2 ) N(\mu,\sigma^2) N ( μ , σ 2 ) Why it feels right: the bell looks the same shape. Fix: tables are only for N ( 0 , 1 ) N(0,1) N ( 0 , 1 ) Always convert with z = ( x − μ ) / σ z=(x-\mu)/\sigma z = ( x − μ ) / σ
Common mistake "Normal's parameter isσ 2 \sigma^2 σ 2 σ 2 \sigma^2 σ 2 σ 2 π \sigma\sqrt{2\pi} σ 2 π Why it feels right: the notation N ( μ , σ 2 ) N(\mu,\sigma^2) N ( μ , σ 2 ) σ 2 \sigma^2 σ 2 Fix: the standard deviation σ \sigma σ σ 2 π \sigma\sqrt{2\pi} σ 2 π z z z
Recall Feynman: explain to a 12-year-oldImagine waiting for popcorn to pop. Uniform is like "the prize is hidden somewhere on this ruler, no clue where." Exponential is "how long till the first pop" — and it doesn't matter how long you've already waited, the next pop is just as surprising. Gamma is "how long till the fifth pop" — five waits stacked together. Normal is the famous bell: if you average lots of random things (like many kids' heights), the average always makes that hill shape. Beta is a dial that only goes from 0 to 1 — great when the thing you're guessing is itself a chance , like "what fraction of free throws will I make?"
Mnemonic Remember the family"U Eat Gamma's Normal Beta" →
U niform (flat), E xponential (1st event, 1 / λ 1/\lambda 1/ λ G amma (k k k α / λ \alpha/\lambda α / λ N ormal (bell, CLT), B eta (proportions in [ 0 , 1 ] [0,1] [ 0 , 1 ] α / ( α + β ) \alpha/(\alpha+\beta) α / ( α + β ) Exp = Gamma(α = 1 \alpha{=}1 α = 1 ; sum of Exps = Gamma ; Uniform = Beta(1,1) .
Central Limit Theorem — why Normal appears everywhere.Poisson Process — generates both Exponential and Gamma.Gamma Function and Beta Function — the integration tools.Moment Generating Functions — slick way to get means/variances.Conjugate Priors in Bayesian Inference — Beta is conjugate to the Binomial.Discrete Distributions — Geometric is the discrete cousin of Exponential.
Uniform U ( a , b ) U(a,b) U ( a , b ) f ( x ) = 1 b − a f(x)=\frac{1}{b-a} f ( x ) = b − a 1 for
a ≤ x ≤ b a\le x\le b a ≤ x ≤ b , else 0.
Uniform variance ( b − a ) 2 12 \frac{(b-a)^2}{12} 12 ( b − a ) 2 (NOT /4).
Exponential PDF and CDF f ( t ) = λ e − λ t f(t)=\lambda e^{-\lambda t} f ( t ) = λ e − λ t ,
F ( t ) = 1 − e − λ t F(t)=1-e^{-\lambda t} F ( t ) = 1 − e − λ t ,
t ≥ 0 t\ge0 t ≥ 0 .
Exponential mean and variance mean
1 / λ 1/\lambda 1/ λ , variance
1 / λ 2 1/\lambda^2 1/ λ 2 .
Memoryless property P ( X > s + t ∣ X > s ) = P ( X > t ) = e − λ t P(X>s+t\mid X>s)=P(X>t)=e^{-\lambda t} P ( X > s + t ∣ X > s ) = P ( X > t ) = e − λ t .
Relation Exp↔Gamma Exp
( λ ) (\lambda) ( λ ) = Gamma
( α = 1 , λ ) (\alpha{=}1,\lambda) ( α = 1 , λ ) ; sum of
k k k Exp = Gamma
( k , λ ) (k,\lambda) ( k , λ ) .
Gamma PDF f ( x ) = λ α Γ ( α ) x α − 1 e − λ x f(x)=\frac{\lambda^\alpha}{\Gamma(\alpha)}x^{\alpha-1}e^{-\lambda x} f ( x ) = Γ ( α ) λ α x α − 1 e − λ x ,
x > 0 x>0 x > 0 .
Gamma mean and variance mean
α / λ \alpha/\lambda α / λ , variance
α / λ 2 \alpha/\lambda^2 α / λ 2 .
Normal PDF 1 σ 2 π exp ( − ( x − μ ) 2 2 σ 2 ) \frac{1}{\sigma\sqrt{2\pi}}\exp(-\frac{(x-\mu)^2}{2\sigma^2}) σ 2 π 1 exp ( − 2 σ 2 ( x − μ ) 2 ) .
Standardize a Normal Z = ( X − μ ) / σ ∼ N ( 0 , 1 ) Z=(X-\mu)/\sigma\sim N(0,1) Z = ( X − μ ) / σ ∼ N ( 0 , 1 ) ,
P ( X ≤ x ) = Φ ( ( x − μ ) / σ ) P(X\le x)=\Phi((x-\mu)/\sigma) P ( X ≤ x ) = Φ (( x − μ ) / σ ) .
68-95-99.7 rule ≈ 68 % \approx68\% ≈ 68% within
1 σ 1\sigma 1 σ ,
95 % 95\% 95% within
2 σ 2\sigma 2 σ ,
99.7 % 99.7\% 99.7% within
3 σ 3\sigma 3 σ .
Beta PDF x α − 1 ( 1 − x ) β − 1 B ( α , β ) \frac{x^{\alpha-1}(1-x)^{\beta-1}}{B(\alpha,\beta)} B ( α , β ) x α − 1 ( 1 − x ) β − 1 on
( 0 , 1 ) (0,1) ( 0 , 1 ) ,
B = Γ ( α ) Γ ( β ) Γ ( α + β ) B=\frac{\Gamma(\alpha)\Gamma(\beta)}{\Gamma(\alpha+\beta)} B = Γ ( α + β ) Γ ( α ) Γ ( β ) .
Beta mean and variance mean
α α + β \frac{\alpha}{\alpha+\beta} α + β α , var
α β ( α + β ) 2 ( α + β + 1 ) \frac{\alpha\beta}{(\alpha+\beta)^2(\alpha+\beta+1)} ( α + β ) 2 ( α + β + 1 ) α β .
Beta(1,1) equals Uniform(0,1).
Gamma function recursion Γ ( α + 1 ) = α Γ ( α ) \Gamma(\alpha+1)=\alpha\Gamma(\alpha) Γ ( α + 1 ) = α Γ ( α ) ,
Γ ( n ) = ( n − 1 ) ! \Gamma(n)=(n-1)! Γ ( n ) = ( n − 1 )! ,
Γ ( 1 / 2 ) = π \Gamma(1/2)=\sqrt\pi Γ ( 1/2 ) = π .
Why 2 π \sqrt{2\pi} 2 π Gaussian integral
∫ e − z 2 / 2 d z = 2 π \int e^{-z^2/2}dz=\sqrt{2\pi} ∫ e − z 2 /2 d z = 2 π (polar-coord proof) normalizes it.
max ignorance flat density
models a random probability
Mean and Variance integrals
Poisson process rate lambda
Intuition Hinglish mein samjho
Intuition Hinglish mein samjhoDekho beta, ye jo continuous distributions hain, inko yaad karne ka smart tarika ye hai ki inhe alag-alag isolated formulas ki tarah mat dekho — inhe ek family tree ki tarah samjho. Har distribution ek alag physical story batati hai. Jaise Uniform matlab total ignorance — pata hai value a a a b b b k k k 0 0 0 1 1 1
Ab sabse zaroori baat — ye jo saari machinery hai (PDF, CDF, mean, variance) wo ek hi baar seekhni padti hai, phir har distribution mein reuse hoti hai. Mean nikalne ke liye bas ∫ x f ( x ) d x \int x\,f(x)\,dx ∫ x f ( x ) d x E [ X 2 ] − μ 2 \mathbb{E}[X^2]-\mu^2 E [ X 2 ] − μ 2 f ( x ) f(x) f ( x )
Ek beautiful cheez dekho — memorylessness ki. Exponential distribution ki khaasiyat ye hai ki agar tum bus ke liye 5 minute wait kar chuke ho, to bhi aage ka wait waisa hi hai jaise abhi start kiya ho. Formula mein P ( X > s + t ∣ X > s ) = e − λ t P(X>s+t \mid X>s) = e^{-\lambda t} P ( X > s + t ∣ X > s ) = e − λ t
