automl工具入门介绍

前言

自动化机器学习已经被广泛应用于各种(跨)业务场景的模型构建，实验以及生产部署当中。
automl领域中有各种各样的开源项目可以直接使用，本篇文章尝试对一些主流的开源框架进行介绍。

• autogluon

• hypergbm

• h2o automl

• lightautoml

• FLAML

备注：本篇文章的所使用的数据集为tabular-playground-series-may-2021数据集.

import pandas as pd
## load data
train_data = pd.read_csv("../input/tabular-playground-series-may-2021/train.csv",index_col=0)
test_data = pd.read_csv("../input/tabular-playground-series-may-2021/test.csv",index_col=0)
target = 'target'
reward_metric = 'auc'

---

---

AutoGluon

AutoGluon: AutoML for Text, Image, and Tabular Data . by Amazon Web Services - Labs

## import packages
from autogluon.tabular import TabularPredictor

model_autogluon = TabularPredictor(label='target')
model_autogluon.fit(train_data=train_data, time_limit=300)

## check leaderboard
model_autogluon.leaderboard()

##predict
preds = model_autogluon.predict(test_data)

运行过程展示:

---

---

Hypergbm

Hypergbm is A full pipeline AutoML tool for tabular data. by DataCanvas

## import packages 
from hypergbm import make_experiment
exp = make_experiment(train_data.copy(),target=target)
estimator = exp.run()

preds = estimator.predict(test_data)

运行过程展示:

---

---

H2O AutoML

H2O is an Open Source, Distributed, Fast & Scalable Machine Learning Platform: Deep Learning, Gradient Boosting (GBM) & XGBoost, Random Forest, Generalized Linear Modeling (GLM with Elastic Net), K-Means, PCA, Generalized Additive Models (GAM), RuleFit, Support Vector Machine (SVM), Stacked Ensembles, Automatic Machine Learning (AutoML). by H2O.ai

## import packages
import h2o
from h2o.automl import H2OAutoML

h2o.init()

h2o_train = h2o.H2OFrame(train_data.copy())
h2o_test = h2o.H2OFrame(test_data.copy())
h2o_train[target] = h2o_train[target].asfactor()

feature_columns = train_data.iloc[:, 1:-1].columns.values
aml = H2OAutoML(max_runtime_secs=300)

aml.train(x=list(feature_columns), 
    	  y=target, 
          training_frame=h2o_train
		  )

aml.leaderboard
aml.predict(h2o.H2OFrame(h2o_test))

运行过程展示:

---

---

Lightauotml

LAMA - automatic model creation framework. by Sberbank AI Lab

## import packages
from lightautoml.automl.presets.tabular_presets import TabularAutoML
from lightautoml.tasks import Task

## run model
model_laml = TabularAutoML(task = Task('multiclass'), timeout = 300)
model_laml.fit_predict(train_data=train_data, roles={'target': 'target'})
model_laml.predict(test_data)

运行过程展示:
默认过程没有输出

---

---

FLAML

Flaml is a fast and lightweight AutoML library. by Microsoft

运行过程展示:

***** 持续更新中