DANA 4830 — Multivariate feature selection Part 2: CON, DISR, and fraud dataset pipeline
Part 2 of multivariate (bivariate-and-beyond) feature selection from Zouhri et al. (2024): CON, DISR, how they differ from CFS, and a full Python pipeline yo...
Recent posts
DANA 4830 — Multivariate feature selection Part 1: mRMR and CFS
Multivariate (sometimes called bivariate-and-beyond) feature selection picks a subset of columns by judging them together. Univariate methods pick strong fea...
DANA 4830 — Univariate feature selection: Chi2, ANOVA, MI, Pearson, ReliefF
Univariate feature selection ranks one column at a time. The paper (Zouhri et al., 2024) uses five univariate filters, then keeps the top 20%, 40%, or 60% of...
DANA 4830 — Feature selection paper: how to read it, MCAR/MAR/NMAR, and the big ideas
Study notes for DANA 4830 on the paper:
DANA 4840 — K-means (exploración 2): USArrests, escala, codo y salida de R
Notas de exploración sobre K-means con datos reales (USArrests): qué hace la función, por qué scale(), cómo leer tamaños, centroides, WSS, betweenss/totss y ...