EvalML is an AutoML library that builds, optimizes, and evaluates machine learning pipelines using domain-specific objective functions.
Project description
EvalML is an AutoML library to build optimized machine learning pipelines for domain-specific objective functions.
Key Functionality
- Domain-specific - Includes repository of domain-specific objective functions and interface to define your own
- End-to-end - Constructs and optimizes pipelines that include imputation, feature selection, and a variety of modeling techniques
- Data Checks - Carefully cross-validates to prevent overfitting and warns you if training and testing results diverge
Install
pip install evalml --index-url https://install.featurelabs.com/<KEY>
Quick Start
Define objective
from evalml import AutoMLSearch
from evalml.objectives import FraudCost
fraud_objective = FraudCost(
retry_percentage=.5,
interchange_fee=.02,
fraud_payout_percentage=.75,
amount_col="amount"
)
Run automl
automl = AutoMLSearch(problem_type='binary', objective=fraud_objective,
max_pipelines=3)
automl.search(X_train, y_train)
See all pipeline ranks
automl.rankings
Get best pipeline and predict on new data
pipeline = automl.best_pipeline
pipeline.predict(X_test)
Next Steps
Read more about EvalML in our Documentation.
Built at Alteryx Innovation Labs
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
evalml-0.11.2.tar.gz
(2.7 MB
view hashes)