Particle Swarm Optimization Search

Installation

# To install the required package, use the following command:
# !pip install modeva

Authentication

# To get authentication, use the following command: (To get full access please replace the token to your own token)
# from modeva.utils.authenticate import authenticate
# authenticate(auth_code='eaaa4301-b140-484c-8e93-f9f633c8bacb')

Import required modules

from modeva import DataSet
from modeva import TestSuite
from modeva.models import MoLGBMClassifier
from modeva.models import ModelTunePSO

Load Dataset

ds = DataSet()
ds.load(name="TaiwanCredit")
ds.set_random_split()

Run PSO search

param_bounds = {"max_depth": [1, 4],
                "learning_rate": [0.01, 1.0]}
param_types = {"max_depth": "int"}

model = MoLGBMClassifier(verbose=-1)
hpo = ModelTunePSO(dataset=ds, model=model)
result = hpo.run(param_bounds=param_bounds,
                 param_types=param_types,
                 n_iter=2,
                 n_particles=10,
                 metric=("AUC", "LogLoss"),
                 cv=5)
result.table

	max_depth	learning_rate	AUC	LogLoss	AUC_rank	LogLoss_rank	mean_fit_time
38	1	0.8343	0.7751	0.4367	1	1	0.1424
18	1	0.8343	0.7751	0.4367	1	1	0.1421
8	1	0.8343	0.7751	0.4367	1	1	0.1413
27	1	0.6666	0.7748	0.4368	4	4	0.1411
32	2	0.6494	0.7706	0.4407	5	8	0.1603
2	2	0.6494	0.7706	0.4407	5	8	0.1599
12	2	0.6494	0.7706	0.4407	5	8	0.1623
17	1	0.0963	0.7696	0.4401	8	5	0.1436
37	1	0.0963	0.7696	0.4401	8	5	0.1413
7	1	0.0963	0.7696	0.4401	8	5	0.1429
15	3	0.5336	0.7655	0.4479	11	11	0.1819
35	3	0.5336	0.7655	0.4479	11	11	0.1807
5	3	0.5336	0.7655	0.4479	11	11	0.1809
11	3	0.5494	0.7641	0.4494	14	14	0.1832
1	3	0.5494	0.7641	0.4494	14	14	0.1811
31	3	0.5494	0.7641	0.4494	14	14	0.1820
14	4	0.3896	0.7641	0.4504	17	20	0.2082
34	4	0.3896	0.7641	0.4504	17	20	0.2148
4	4	0.3896	0.7641	0.4504	17	20	0.2063
33	2	0.8929	0.7633	0.4497	20	17	0.1595
3	2	0.8929	0.7633	0.4497	20	17	0.1603
13	2	0.8929	0.7633	0.4497	20	17	0.1617
26	2	1.0000	0.7632	0.4510	23	23	0.1601
22	2	1.0000	0.7632	0.4510	23	23	0.1619
23	2	1.0000	0.7632	0.4510	23	23	0.1609
25	2	1.0000	0.7632	0.4510	23	23	0.1600
28	2	1.0000	0.7632	0.4510	23	23	0.1601
24	3	0.7842	0.7559	0.4676	28	31	0.1803
0	3	0.7180	0.7557	0.4634	29	28	0.1802
10	3	0.7180	0.7557	0.4634	29	28	0.1816
30	3	0.7180	0.7557	0.4634	29	28	0.1802
19	3	0.8713	0.7512	0.4758	32	32	0.1820
9	3	0.8713	0.7512	0.4758	32	32	0.1805
39	3	0.8713	0.7512	0.4758	32	32	0.1791
21	3	1.0000	0.7464	0.4915	35	38	0.1823
20	3	1.0000	0.7464	0.4915	35	38	0.1830
16	3	0.9263	0.7433	0.4889	37	35	0.1820
36	3	0.9263	0.7433	0.4889	37	35	0.1796
6	3	0.9263	0.7433	0.4889	37	35	0.1822
29	4	1.0000	0.7256	0.5610	40	40	0.2073

result.plot("parallel", figsize=(8, 6))

result.plot(("max_depth", "AUC"))

result.plot(("learning_rate", "AUC"))

Retrain model with best hyperparameter

model_tuned = MoLGBMClassifier(**result.value["params"][0],
                             name="LGBM-Tuned",
                             verbose=-1)
model_tuned.fit(ds.train_x, ds.train_y)
model_tuned

MoLGBMClassifier(boosting_type='gbdt', class_weight=None, colsample_bytree=1.0,
                 importance_type='split', learning_rate=0.7180374727086953,
                 max_depth=3, min_child_samples=20, min_child_weight=0.001,
                 min_split_gain=0.0, n_estimators=100, n_jobs=None,
                 num_leaves=31, objective=None, random_state=None,
                 reg_alpha=0.0, reg_lambda=0.0, subsample=1.0,
                 subsample_for_bin=200000, subsample_freq=0, verbose=-1)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Diagnose the tuned model

ts = TestSuite(ds, model_tuned)
result = ts.diagnose_accuracy_table()
result.table

	AUC	ACC	F1	LogLoss	Precision	Recall	Brier
train	0.8555	0.8412	0.5539	0.3697	0.7385	0.4432	0.1157
test	0.7646	0.8158	0.4633	0.4441	0.6260	0.3678	0.1379
GAP	-0.0909	-0.0254	-0.0906	0.0744	-0.1125	-0.0754	0.0222