Learn FLAML’s budget-aware AutoML workflow: define task/metric/time budget, fit classification or regression models, control estimator scope, inspect the best configuration and deliver a validated model.
Enter your name. Complete at least 12 practice cases to unlock the Certificate of Participation.
FLAML focuses on finding strong configurations efficiently under explicit resource constraints instead of exhaustively searching every possibility.
A realistic budget turns AutoML into an operational planning tool instead of an open-ended experiment.
from flaml import AutoML
automl = AutoML()A realistic budget turns AutoML into an operational planning tool instead of an open-ended experiment.
For classification, pass X/y, set task="classification", choose an appropriate metric and provide a time budget in seconds.
The short API is convenient, but task and metric selection still determine what FLAML optimizes.
automl.fit(
X_train=X_train, y_train=y_train,
task='classification',
metric='roc_auc',
time_budget=60
)The short API is convenient, but task and metric selection still determine what FLAML optimizes.
FLAML can search regression estimators under the same budget-aware framework with regression metrics such as MAE, MSE or R2.
A budget-aware search is useful only if the objective metric reflects the real cost of prediction errors.
automl.fit(
X_train=X_train, y_train=y_train,
task='regression',
metric='mae',
time_budget=60
)A budget-aware search is useful only if the objective metric reflects the real cost of prediction errors.
The time_budget setting defines how long FLAML can search; increasing it can improve search depth but also changes operational cost.
Time is part of the experiment definition; record it with the score.
settings = {
'time_budget': 120,
'metric': 'accuracy',
'task': 'classification'
}
automl.fit(X_train=X_train, y_train=y_train, **settings)Time is part of the experiment definition; record it with the score.
When deployment rules or libraries restrict acceptable model families, constrain estimator_list so search reflects what you can actually ship.
A constrained search space can be more useful than a broader one that returns models you cannot deploy.
automl.fit(
X_train=X_train, y_train=y_train,
task='classification',
time_budget=60,
estimator_list=['lgbm','xgboost','rf']
)A constrained search space can be more useful than a broader one that returns models you cannot deploy.
After search, inspect the selected estimator, configuration and training history instead of treating the result as a black box.
The best configuration is useful evidence; the final acceptance decision still belongs to independent validation.
print(automl.model.estimator)
print(automl.best_config)The best configuration is useful evidence; the final acceptance decision still belongs to independent validation.
FLAML AutoML follows estimator-like prediction methods, making it straightforward to score labels or probabilities after the search.
Probability outputs are not automatically calibrated for every use case; validate before threshold-based decisions.
y_pred = automl.predict(X_test)
y_prob = automl.predict_proba(X_test)Probability outputs are not automatically calibrated for every use case; validate before threshold-based decisions.
A fast AutoML result still needs data-quality checks, reproducibility, deployment testing, monitoring and a retraining policy.
FLAML helps you spend the search budget wisely; governance decides whether the result is safe and useful to ship.
# Search → independent test → package → monitor
print(automl.model.estimator)FLAML helps you spend the search budget wisely; governance decides whether the result is safe and useful to ship.
Open each item only after answering it in your own words.
Strong model performance while using search resources efficiently under explicit constraints.
The wall-clock search budget, in seconds, for AutoML fitting.
To align the search with approved, deployable model families and operational constraints.
Which underlying estimator FLAML selected as the current best model.
Independent testing, deployment validation, monitoring and retraining governance.
| Need | Tool | Focus |
|---|---|---|
| Low-code end-to-end experiment workflow | PyCaret | |
| Scikit-learn pipeline search + ensembles | auto-sklearn | |
| Scalable platform + leaderboard + stacked ensembles | H2O AutoML | |
| Evolutionary pipeline structure search | TPOT | |
| Hyperparameter optimization for your chosen model/code | Optuna | |
| Budget-aware fast AutoML search | FLAML | ★ Current training |
Choose from the problem, validation evidence, compute budget and delivery constraints—not from popularity alone.
The technical concepts and code patterns in this training follow the project’s official documentation. Validate package versions and environment compatibility before production use.
“I can run FLAML classification/regression searches under explicit time budgets, choose defensible metrics, constrain estimator families, inspect the selected model/configuration and validate it independently before deployment.”
Complete at least 12 of the 24 practice cases (50%) and enter your name.