Underfitting
noun
say it: “UN-der-fit-ing” /ˈʌndərˌfɪtɪŋ/
When a model is too simple to learn the real pattern in its data, so it predicts poorly even on the examples it learned from.
What is underfitting?
Underfitting is when a model is too simple to catch the pattern, such as a straight line through harvests that rise and then fall. It scores badly on its training data and on new data alike; the fix is a more flexible model or better features.
from sklearn.linear_model import LinearRegression
rain = [[10], [20], [30], [40], [50]]
apples = [100, 160, 180, 160, 100] # up, then down
line = LinearRegression().fit(rain, apples)
print(round(line.score(rain, apples), 2)) # 0.0: no better than guessing the average
In a sentence
- “Adding polynomial features cured the underfitting.”
- “The straight line was underfitting: it missed even the training orchards.”
Level: Journeyman · More AI Basics terms at this level
Updated