Topic 6 · Functions, Exponential and Logarithmic Functions
[video] Lecture video — Topic 6: Functions, Exponential and Logarithmic Functions — /api/lesson-media/videos/4d3653ff-a3ca-4f63-bf8e-bb93c553e599-1794892659738-0f2b09f572614890.mp4
Flashcards · Functions, exponentials and logarithms
Function, domain, range — A function assigns exactly one output to each input in its domain. The range is the set of outputs actually taken.
f(x) = x² on ℝ has range [0, ∞).
Natural real domain — The real inputs for which a formula is defined.
ln(x − 2) needs x − 2 > 0, so the domain is (2, ∞).
Affine function — f(x) = b + wx: intercept b and constant slope w, so f(x + h) − f(x) = wh at every x.
In ML, b + wx is called a linear model.
Exponential and its range — eˣ has domain ℝ and range (0, ∞): always positive, never 0.
3 + e^(−x) has range (3, ∞).
Natural logarithm — ln is the inverse of eˣ: ln y = x ⇔ eˣ = y for y > 0. Also ln(eˣ) = x, ln 1 = 0, ln e = 1.
Three log rules — ln(uv) = ln u + ln v; ln(u/v) = ln u − ln v; ln(uᶜ) = c ln u. There is no rule for ln(u + v).
ln(2e³) = ln 2 + 3.
Exponential growth on the log scale — A(t) = A₀e^(kt) gives ln A(t) = ln A₀ + kt, a straight line with slope k. Doubling time = ln 2 / k.
A rise of 10% per year has k = ln 1.1 ≈ 0.0953.
Sigmoid — σ(z) = 1/(1 + e^(−z)) maps any score to a probability strictly between 0 and 1; σ(0) = 1/2 and σ is increasing.
Odds and log-odds — odds = p/(1 − p). The log-odds ln(p/(1 − p)) is the inverse of σ. One unit of score multiplies the odds by e.
p = 3/4 has odds 3 and score ln 3.
[markdown] content available only on the platform
