Odds measure an event’s probability relative to the probability of its complement: the ratio of success to failure. In classification, they express how strongly a prediction favors one outcome over the other. section 5.1.3
Logistic regression models log-odds before the sigmoid function converts them into probabilities. A positive feature weight raises the odds that a molecule is a CYP3A4 inhibitor; a negative weight lowers them, with other features fixed. section 5.1.3
Interpreting a feature weight
If a feature has weight 0.45, increasing it by one unit multiplies the predicted odds by exp(0.45) = 1.57—a 57% increase in inhibitor odds, assuming all other features remain unchanged. section 5.1.3
The odds ratio for increasing feature xj by one unit is exp(wj), directly linking a logistic-regression weight to multiplicative change in predicted odds. section 5.1.3




