Définition
Automated machine learning, or AutoML, lowers the barrier to entry for creating models. It can select an algorithm, prepare data, tune hyperparameters, compare options, and produce a finished model. This is useful for analysts and teams who need results without manually going through dozens of settings.
Exemple
The analyst loads the sales table, selects the forecast goal, and the AutoML system itself compares several models and offers the best one based on metrics.
Pourquoi c'est important
The term is important for business: AutoML allows you to quickly test hypotheses, but does not replace the understanding of the data and the quality of the result.
Fonctionnement
The platform builds a cycle of experiments: data processing, feature selection, model training, parameter selection, evaluation and sometimes deployment.
Où c'est utilisé
- forecasting
- tabular data
- quick experiments with models
Limites
AutoML may provide a formally good model, but lack understanding of the business context, data leaks, or the wrong goal. The results must be verified by an expert.
FAQ
Why is “AutoML” useful to know?
The term is important for business: AutoML allows you to quickly test hypotheses, but does not replace the understanding of the data and the quality of the result.