Todos os livros

Profissional

Apps Sobre Coach Entrar Começar a ler

Quantitative Finance · Glossário

O que é Label concurrency, average uniqueness?

Também chamado de: label concurrency · average uniqueness

Definition 2.4 Machine Learning for Markets · Capítulo 2 — Targets, Labels and Sample Weights

Label ii spans the bars (t0,i,t1,i](t_{0,i}, t_{1,i}]. The label concurrency ctc_t of bar tt is the number of labels whose span contains it. The average uniqueness of label ii is uˉi=1t1,i−t0,i∑t∈(t0,i,t1,i]1/ct\bar u_i = \frac{1}{t_{1,i} - t_{0,i}}\sum_{t\in(t_{0,i}, t_{1,i}]}1/c_t: the share of its information it holds alone.

Concurrency and average uniqueness on seven bars: labels A and B share bars 2 and 3; C is alone. The three labels are worth 0.67 + 0.67 + 1 = 2.33 independent labels, the number of bars they cover divided by their length only when all lengths are equal ().
Figure 2.3. Concurrency and average uniqueness on seven bars: labels A and B share bars 2 and 3; C is alone. The three labels are worth 0.67+0.67+1=2.330.67 + 0.67 + 1 = 2.33 independent labels, the number of bars they cover divided by their length only when all lengths are equal (Proposition 2.5).
Ler no capítulo →