Multi-Feature Likelihood-Ratio Screening of Pre-Owned Luxury Goods: Fusing Stitch Pitch, Hardware Spectra and Leather Grain, and the Bayesian Boundary between Lab Referral and Instant Purchase: chapter answers
At the intake desk of a pre-owned luxury store a belief prevails: the more one looks at stitching, hardware, spectra and leather grain, the surer one can be. This book refutes the default premise behind that belief, that these feature readings are independent pieces of evidence. Within one class of items they often share common factors such as wear, photographic lighting, operator habits and batches of parts, so when naive Bayes multiplies their likelihood ratios it records the same evidence several times. Every probability in the book is an appearance-probability reference only, not an offic…
DOI pending · monograph page goes live after publication · 5DL-B-281 · 中文版
- Chapter 1 · Weight of Evidence: Single-Feature Likelihood Ratios, Decibans and the Sum of Four Features
A feature yields a likelihood ratio, its logarithm yields an additive weight of evidence, and the deciban is a convenient unit. With the illustrative parameters, the neutral point for stitch pitch lies at 0.60 spreads,… - Chapter 2 · Double Counting: How Correlated Features Make Naive Bayes Overconfident
Chapter 2 is the core: with illustrative parameters, items the naive posterior scores above 0.99, at an average of 0.996, are non-conforming in a share of only 0.756; the inflation grows with within-class correlation an… - Chapter 3 · Correcting for Correlation: the Effective Number of Independent Features, Exact Weights and a One-Number Slope
Chapter 3 gives the effective number of independent features (four features are worth only 2.007 under the illustrative matrix) and a one-number slope of 0.571. - Chapter 4 · A Loss Matrix and Three Zones: Instant Buy, Referral and Decline
Chapter 4 derives three zones from a loss matrix: an instant buy below a posterior of 0.0539, a referral from 0.0539 to 0.8362, and a decline above; by Good's theorem the gross value of free information is never negativ… - Chapter 5 · Measuring More or Stopping: Cheap Evidence Is Worth Most at the Boundaries
Chapter 5 shows that cheap extra measurements are worth most at the boundaries (an independent one is worth 444.4 yuan on the instant-buy boundary but only 18.2 yuan in the middle of the band) and that sequential measur… - Chapter 6 · Channel Priors, Audits and Calibration: What Gets Referred Decides What a Store Can Learn
The prior decides which zone the same evidence falls into. With the illustrative parameters, evidence of 4.0 decibans toward N leads to an instant buy in the repeat-consignor channel and to a referral in the private-sel… - Chapter 7 · East and West in the Sources: Suspecting the Neighbour of the Axe, and Peirce's Thermometer of Belief
Chapter 7 examines the suspected neighbour of the Liezi and Peirce's log odds of 1878. - Chapter 8 · Closing the Loop: a Three-Zone Intake Procedure, the Refer-or-Buy Screener and a Pre-Registered Test
Chapter 8 gives a three-zone intake procedure, the Refer-or-Buy Intake Screener and a pre-registered test plan: testing within-class correlation needs 194 authenticated conforming items, testing log loss 373, and testin…
Illustrative model calculation, not a live quote