Une distribution décrit la fréquence, l’étendue et la probabilité des valeurs possibles d’une caractéristique, d’une étiquette, d’une variable ou d’une sortie de modèle. Comprendre les distributions est essentiel pour l’analyse des données, la détection de dérive, l’évaluation des modèles et l’évaluation de l’incertitude.
The frequency and range of different values for a given feature or label. A distribution captures how likely a particular value is. The following image shows histograms of two different distributions: - On the left, a power law distribution of wealth versus the number of people possessing that wealth. - On the right, a normal distribution of height versus the number of people possessing that height. !Two histograms. One histogram shows a power law distribution with wealth on the x-axis and number of people having that wealth on the y-axis. Most people have very little wealth, and a few people have a lot of wealth. The other histogram shows a normal distribution with height on the x-axis and number of people having that height on the y-axis. Most people are clustered somewhere near the mean. Understanding each feature and label's distribution can help you determine how to normalize values and detect outliers. The phrase out of distribution refers to a value that doesn't appear in the dataset or is very rare. For example, an image of the planet Saturn would be considered out of distribution for a dataset consisting of cat images.