PDFs & probability as area
A continuous variable has no probability table — it has a curve , and probability = area under that curve. There is no to read off (a single point has no width, so that area is 0); every question is “shade the region, find its area” — by geometry when is straight lines, otherwise by integrating.
Probability is the area under f(x)
A valid PDF has and total area ; then (integrate only the slice asked for). Since , and are interchangeable.
Probability = area. Total area = 1; a probability is the area of one slice; a single point has probability 0, so < and ≤ are the same.
Worked example
Easiest first: probability as a shape (geometry)
When every piece of is a straight line you can skip calculus and read the area off as triangles, trapezia and rectangles. For the taxi pdf above, is a trapezium plus a rectangle :
Same answer as integrating each piece — geometry is the faster route whenever is linear.
Straight-line f? Use geometry (½·base·height, trapezium, rectangle). Curved f? Integrate. Either way you are just finding an area.
Worked example
Find a constant k, then a tail (calculus)
Water erupts to height m with for . Set the total area to 1, then read off the tail.
- Total area: .
- Tail: .
“Show that” / find a constant: set the TOTAL area = 1 and solve exactly (fractions, not decimals).
Worked example
An infinite upper limit
A battery lasts years with for . An limit is no harder: substitute it and the term goes to 0.
- It is a PDF: .
- Tail: .
An ∞ limit just means “substitute infinity”: 1/x, 1/x² and e^(−λx) all vanish, leaving only the finite (lower) term.
Worked example
A two-piece pdf: split, then add to 1
on and on . Integrate each piece, add, set to 1.
- A constant / “show that”: set the total area .
- A probability : integrate (or take the area of) only that slice.
- Piecewise? split at each breakpoint and add — integrate only the parts inside the interval.
| Asked | Do |
|---|---|
| find k / 'show that' | ∫f over the whole domain = 1 |
| P(a<X<b) | ∫ from a to b only (or area by geometry) |
| median / quartile | area from the left = 0.5 / 0.25 |
| E(X) | ∫ x·f |
| Var(X) | ∫ x²·f − [E(X)]² |
Examiner note
Common mistake