Underfitted

Part of speech: adjective, verb

Definitions

1. Describing a model or system that has insufficient complexity to capture the underlying pattern of data, resulting in poor performance and inaccurate predictions
2. Characterizing an analytical approach or algorithm that fails to adequately represent the variability in data due to overly simplistic assumptions or insufficient training
3. Referring to a statistical or machine learning model that is too simple to fit the data well, leading to underperformance and inability to generalize effectively

Etymology: "Underfitted" is the past tense and past participle of "underfit," a verb meaning to fit a model too simply, so that it fails to capture the pattern in its data. The verb is formed from the prefix "under-," meaning insufficient or below the proper amount, and "fit," the ordinary English verb for making something match or conform to a shape or need. Put together, the sense is literally "fitted too little." The verb's current, specialized sense comes from statistics and machine learning, where "fit" refers to adjusting a model's parameters so it matches observed data. A model that has been underfitted is too rigid or too simple to follow real structure, the mirror image of overfitting, where a model memorizes noise. The word's base components are old and everyday, so the technical meaning is a modern extension of familiar English rather than a borrowing from another language.

Synonyms: undertrained, underdeveloped, underoptimized

Antonyms: overfitted, overtrained, optimized, well-fitted