Each dot is a house in Tamil Nadu; its colour = its price/sq-ft. We can't survey all houses (the population) — so we draw a random sample and estimate. Watch how a bigger sample lands closer to the truth.
Sample size:
Population avg (truth)
—
Sample avg (estimate)
—
Error |est − truth|
—
Population (all houses)Selected in sampleLower bar = sampling distribution of the estimate (it gets tighter with bigger n)
Try this: Keep "Draw a Sample" on size 10 and watch the estimate jump around. Switch to 100 — the jumps shrink. That's the core idea of inferential statistics: one spoon (sample) predicts the whole pot (population), and a bigger spoon predicts better.
Hinglish: Poora patila = Population. Ek chammach = Sample. Bada chammach (n=100) → taste ka andaza zyada sahi. 🍛
📏 Scales of Measurement
Pick a variable. Watch which operations unlock as you climb the staircase. Each higher scale keeps everything below it and adds one new power.
👕 Jersey number🩸 Blood group🏁 Race rank (1st,2nd,3rd)⭐ Service rating🌡️ Temp (°C)📝 Exam marks📐 Height (cm)
1 · Nominal
Names / labels only — no order
=≠
2 · Ordinal
Labels + meaningful order/rank
=≠<>
3 · Interval
Order + equal gaps, but no true zero
=≠<>+−
4 · Ratio
Equal gaps + a true zero → ratios work
=≠<>+−×÷
🗓️ Time-Series vs Cross-Sectional
Same temperature table — two ways to slice it. A row = one city over many days (time-series). A column = many cities on one day (cross-sectional).