AI glossary: terms starting with H

all the points on one side of a line or plane

Example: the positive side

The number of positions at which two rows differ.

Catch an exception with except, so the program carries on.

A number Python computes from a value, which a set uses to decide where the value is stored. A dictionary uses its keys' hashes in the same way.

A function that turns a value into an int, the same int every time for the same value. Python's is hash().

A list of buckets, where a value's hash decides its bucket, so a lookup compares only with the keys in one bucket. Every Python set and dict is one.

A value whose hash stays stable during its lifetime and whose equal values have equal hashes, so it can be a set member or dictionary key: numbers, strings, and tuples of hashable values. Lists, sets and dictionaries are not.

a value a model produced, an estimate

Example: ŷ = (5, −3, 10)

The matrix that turns y into the fitted values; its diagonal is the leverage.

A matrix whose cell color encodes its numeric value.

More or larger high values than the chosen reference predicts.

Read with every other feature in the model unchanged.

Labelled examples kept out of training, used only to score the model: 1,115 messages.

Prediction error measured on rows withheld from fitting.

the matrix of all second partial derivatives

Example: diag(2, 6) for x² + 3y²

Error spread changes with predictor value.

An equal-area cell that counts paired observations in a two-variable plot.

Written in words and symbols such as print, if and +, while Python handles the details of the computer's memory and hardware.

Training and validation curves that meet at a low score.

A training curve well above the validation curve.

A function that takes a function as an argument, or returns one: sorted with key=, map, filter.

Counts observations inside adjacent numeric intervals.

Counts of values in bins; with density on the vertical axis, each class's bars have the same total area, so a larger class does not look more important.

Rows reserved from model-directed exploration for later evaluation: the 6,000 pinned IDs.

Squared inside a threshold and linear outside it.

People deciding the label of each example.

every component inside the same interval

Example: the unit cube [0, 1]ᵈ

A setting of a model chosen before training, not learned in fit: k in KNNRegressor(k=2).

the points where w · x + b = 0, in any dimension |

every point at distance r from a centre, in d dimensions

Example: ‖x − c‖₂ = r