ZJUEarthData / Geochemistrypi

an open-sourced highly automated machine learning Python framework for data-driven geochemistry discovery
https://geochemistrypi.readthedocs.io/en/latest/
MIT License
83 stars 29 forks source link

feat: Adding AdaBoost Classification #402

Open HaibinLai opened 1 week ago

HaibinLai commented 1 week ago

Added AdaBoost Classification with FlAML framework.

Testing

Choosing dataset and Classification ML workflow

Select data: d66255f84c9bfce4410cdbcbf4c6c186

Here we select column 2 "Label" for Y set, column [3,12] as X set. 9fa29828e8490f6a1f8c1ce66220e025

Then choose mode2: Classification. Here we can see "AdaBoost" in label 11: 9de4f361e67ae3c786edade1b95eed9e

In feature engineering part we get the data. 4349bd32d6a2fefa531a7fe5b63628d1

AdaBoost in non-autoML workflow

Select parameters for AdaBoost. Here we expose 3 hyperparameter in the model: N estimator for how many decision trees we are going to use, Learning rate for the model, max_depth for the depth of the decision trees used in AdaBoost.

4477471d0b6bd9f34de78579ba3cc823

Training and collecting result:

b9b146441a0ed6e5711e757fc5d7c7de

d8881aee06faa8221cfa89a4c98b28fe

AutoML Workflow

Select AutoML here: image

Auto tuning: image image

image

Final Result: 392a86a04e416eed135a71c83a1089d5

HaibinLai commented 1 week ago

Choosing dataset and Classification ML workflow

image image image

AdaBoost in non-autoML workflow

image image image image image image

AutoML image image image

final Result image