lunax 0.0.6

A machine learning framework.

CN | EN

Lunax is a machine learning framework specifically designed for the processing and analysis of tabular data. The name Lunax is derived from the name of a beloved feline mascot at South China University of Technology 猸愶笍 Star it if you like it 猸愶笍

---

Installation

conda create -n your_env_name python=3.11
conda activate your_env_name
pip install lunax

Features

• Data loading and Data pre-processing
• EDA analysis
• Supports multi-model training and Hyperparameter tuning
• Comprehensive model evaluation and Explainable AI (XAI)

Quick Start

Data Loading and Pre-processing

from lunax.data_processing.utils import *
df_train = load_data('train.csv') # or df = load_data('train.parquet',mode='parquet')
target = 'label_column_name'
df_train = preprocess_data(df_train,target) # data pre-processing, including missing value handling, feature encoding, feature scaling
X_train, X_val, y_train, y_val = split_data(df_train, target)

Exploratory Data Analysis

from lunax.viz import numeric_eda, categoric_eda
numeric_eda([df_train,df_test],['train','test'],target=target) # numeric feature analysis
categoric_eda([df_train,df_test],['train','test'],target=target) # categorical feature analysis

Automation Machine Learning Modeling

from lunax.models import xgb_clf # or xgb_reg, lgbm_reg, lgbm_clf, cat_clf, cat_reg
from lunax.hyper_opt import OptunaTuner
tuner = OptunaTuner(n_trials=10,model_class="XGBClassifier") # Hyperparameter optimizer, n_trials is the number of optimization times
# or "XGBRegressor", "LGBMRegressor", "LGBMClassifier" , "CatClassifier", "CatRegressor"
results = tuner.optimize(X_train, y_train, X_val, y_val)
best_params = results['best_params']
model = xgb_clf(best_params)
model.fit(X_train, y_train)

Model Evaluation and Explainable AI (XAI)

model.evaluate(X_val, y_val)