sampling

Sampling, Validation and ML Diagnostics

Indicator matrix and sequential bootstrap tooling.

Why This Module Exists

Produces less correlated training samples when labels overlap heavily in time.

Key Public APIs

get_ind_matrix
seq_bootstrap
get_ind_mat_average_uniqueness
num_concurrent_events

Core Math

Average Uniqueness

\[u_i=\frac{1}{|T_i|}\sum_{t\in T_i}\frac{1}{c_t}\]

Sequential Draw Prob

\[P(i)\propto E[u_i \mid \mathcal{S}]\]

Code Examples

Run sequential bootstrap

use openquant::sampling::seq_bootstrap;

let ind = vec![vec![1,0,1], vec![0,1,1], vec![1,1,0]];
let idx = seq_bootstrap(&ind, Some(3), None);

Implementation Notes

Indicator matrix quality drives bootstrap quality.
Use average uniqueness as a diagnostics KPI.