2.7.2 · D3Statistics & Probability — Intermediate

Worked examples — Cumulative frequency — ogive, median from graph

3,354 words15 min readBack to topic

The scenario matrix

Every median-from-CF problem is one (or a blend) of these cells:

Cell What makes it special Trap it sets
A lands exactly on a class boundary (CF equals ) Off-by-one median class
B lands mid-class — real interpolation Rounding / fraction slips
C Discontinuous classes () Forget boundary fix
D Twin-ogive — read median off the crossing Reading wrong axis
E Unequal class widths ( differs per class) Use wrong
F Missing frequency given the median Solve two equations together
G Degenerate: median in first / last class, open-ended interval, all data in one class ; open-end boundaries
H Real-world word problem (waiting times, incomes) Extract the table first

We now hit each cell at least once across 8 worked examples (Cell G is split into G(i) and G(ii)).


Setup we reuse


Example 1 — Cell A · on a boundary

  1. . Why this step? For grouped data we split by height , never — the ogive is a smooth curve, not a discrete list.
  2. Find the median class. CF goes from (below 20) up to within the 20–30 class. So median class . Why this step? The median class is where CF crosses or reaches from below — CF was at , then climbs to across this class. It is NOT 30–40.
  3. Extract: . Why this step? is the CF of the class before 20–30 (that's the 10–20 row, CF ).
  4. Interpolate: Why this step? We needed more counts; the class holds exactly , so we travel the whole width — landing at the far edge, .
Figure — Cumulative frequency — ogive, median from graph

Example 2 — Cell B · interpolation genuinely kicks in

  1. . Why? Same halving rule.
  2. Median class. CF jumps inside 20–30, so it contains the 20th count. Median class . Why? — the crossing happens inside this class.
  3. Extract: .
  4. Interpolate: Why? We are counts into a class of ; assuming the 12 values are spread evenly across width 10, we go of the way in. Per our convention we round the final line to .

Example 3 — Cell C · discontinuous classes

  1. Convert to continuous boundaries. Subtract from each lower, add to each upper: Why this step? Real ages between 19 and 20 exist; the true boundary is the midpoint . Now cleanly and no gaps.
  2. CF table: .
  3. . CF crosses going in the class. Median class .
  4. Extract: .
  5. Interpolate: Why? counts into a class of , width 10; rounded to two decimals.

Example 4 — Cell D · twin-ogive (read the crossing)

Forecast: Both curves cross at . Guess: the there should match from Example 2.

  1. Less-than CF: plotted at upper boundaries . Why upper? Less-than CF counts everything below the upper boundary.
  2. More-than CF: plotted at lower boundaries . Why lower? More-than CF at a lower boundary everything from that boundary rightward.
  3. They cross where "number below" "number above" . Drop a perpendicular → . Why this step? Both halves equal only at the median — by definition of the median.
Figure — Cumulative frequency — ogive, median from graph

Example 5 — Cell E · unequal class widths

  1. . CF crosses going in the 30–40 class.
  2. Extract using THIS class's width: . Why this step? The formula's is the width of the median class only — here , even though the class before it had width . Using a neighbour's width is the Cell-E trap.
  3. Interpolate: Why? counts into a class of , width 10; rounded to two decimals.

Example 6 — Cell F · solve for a missing frequency (the honest algebra)

  1. Equation 1 — the total constraint. Why this step? All frequencies must sum to ; with two blanks this is one equation in two unknowns, so we cannot finish on it alone — we need the median equation too.
  2. Locate the median class from the given median. The stated median , so the median class is 30–40. Why this step? The median always lies inside its class; being told it equals pins the class to the one straddling .
  3. Extract the five numbers (median class 30–40). Why this step? is the running total of the three classes before 30–40; is the unknown frequency of the median class itself. Consistency check: for the median to sit inside 30–40 we need ; here ✓, so the problem does have a solution — no "misprint" surgery required.
  4. Equation 2 — the median formula. Substitute into the one formula and set it equal to : Why this step? Everything except is known, so the median equation becomes a single equation in .
  5. Solve for . Why this step? Multiply both sides by , then divide by — pure algebra, clean whole number.
  6. Use Equation 1 to finish. and , so Why this step? The total constraint delivers the second unknown once the median has pinned the first.

Example 7 — Cell G · degenerate cases

  1. . CF crosses inside 0–5 (CF goes ). Median class .
  2. Extract: . Why ? There is no class before the first, so nothing has accumulated — the running total starts at . This is the degenerate boundary case.
  3. Interpolate: Why this step? We need counts out of the living in this class; those are spread evenly across width , so we travel of the width. With the whole median is that step.
  1. . CF crosses going in the 20–40 class. Median class . Why this step? — the crossing is inside 20–40, nowhere near the open top class.
  2. Extract: . Why this step? Every one of the five numbers lives in a closed class (20–40 and its predecessor), so the missing upper edge of "60+" never enters the formula.
  3. Interpolate: Why this step? counts into a class of , width .

Example 8 — Cell H · real-world word problem

  1. Translate the words. "The time below which half the customers fall" the median. Why this step? Word problems hide the table; the phrase "half fall below" is the median's definition.
  2. CF: . .
  3. Median class: CF crosses going in 4–6. Extract .
  4. Interpolate: Why? counts into a class of , width ; rounded to two decimals.

Recall

Recall Match the cell to the trap

First class median ::: (nothing accumulated yet) Discontinuous classes ::: subtract/add before finding on a boundary ::: median class is the one CF crosses from below, not the next one Unequal widths ::: use the median class's own Missing frequency + given median ::: write the total equation AND the median equation, solve together Missing frequency has no solution when ::: (classes before already reach halfway) Open-ended interval ::: harmless if the median class is closed; else assume neighbour's Twin ogive ::: median is the where the two curves cross at height