smallaxe 0.6.3

A PySpark MLOps library for simplified model training and optimization

smallaxe

A PySpark MLOps library that simplifies model training, evaluation, and optimization for PySpark DataFrames.

Why smallaxe?

PySpark MLlib has a steep learning curve and verbose API. smallaxe provides a clean, scikit-learn-like interface for common ML workflows while leveraging the distributed power of Spark.

Features

• Simple API - Train models with familiar fit()/predict() patterns
• Multiple Algorithms - XGBoost, LightGBM, CatBoost, and Random Forest
• Preprocessing Pipeline - Imputer, Scaler, Encoder with chainable pipelines
• Hyperparameter Optimization - Built-in hyperopt integration with early stopping
• Automated Training - Train all algorithms and compare with one call
• Visualization - Plotly-based charts for model evaluation
• Cross-Validation - Train/test split and k-fold with stratified sampling

Installation

pip install smallaxe

Install with optional algorithm dependencies:

pip install smallaxe[xgboost]    # XGBoost support
pip install smallaxe[lightgbm]   # LightGBM support
pip install smallaxe[catboost]   # CatBoost support
pip install smallaxe[all]        # All algorithms

Quick Start

from smallaxe.training import Regressors
from smallaxe.datasets import load_sample_regression

# Load sample data
df = load_sample_regression(spark)

# Train a model
model = Regressors.random_forest()
model.fit(df, label_col='price', exclude_cols=['id'])

# Make predictions
predictions = model.predict(df)

Usage Examples

Training with Cross-Validation

from smallaxe.training import Classifiers

model = Classifiers.xgboost(task='binary')
model.fit(
    df,
    label_col='churn',
    validation='kfold',
    n_folds=5,
    stratified=True
)

print(model.validation_scores)

Preprocessing Pipeline

from smallaxe.pipeline import Pipeline
from smallaxe.preprocessing import Imputer, Scaler, Encoder
from smallaxe.training import Regressors

pipeline = Pipeline([
    ('imputer', Imputer(numerical_strategy='median')),
    ('scaler', Scaler(method='standard')),
    ('encoder', Encoder(method='onehot')),
    ('model', Regressors.xgboost())
])

pipeline.fit(
    df,
    label_col='target',
    numerical_cols=['age', 'income'],
    categorical_cols=['city', 'category']
)

predictions = pipeline.predict(new_df)

Hyperparameter Optimization

from smallaxe.search import optimize
from hyperopt import hp

param_grid = {
    'max_depth': hp.choice('max_depth', [3, 5, 7, 10]),
    'learning_rate': hp.uniform('learning_rate', 0.01, 0.3)
}

best_model = optimize.run(
    model=Regressors.xgboost(),
    dataframe=df,
    label_col='target',
    param_grid=param_grid,
    metric='rmse',
    max_evals=50
)

print(best_model.best_params)

Automated Training

from smallaxe.auto import AutomatedTraining

auto = AutomatedTraining(model_type='classification', metrics=['f1_score', 'auc_roc'])
auto.fit(
    df,
    label_col='churn',
    numerical_cols=['tenure', 'monthly_charges'],
    categorical_cols=['contract'],
    n_folds=5
)

# Compare all models
auto.metrics.show()

# Use best model
predictions = auto.predict(new_df)

Supported Algorithms

Algorithm	Regressor	Classifier	Dependencies
Random Forest	Yes	Yes	None (native PySpark)
XGBoost	Yes	Yes	smallaxe[xgboost]
LightGBM	Yes	Yes	smallaxe[lightgbm]
CatBoost	Yes	Yes	smallaxe[catboost]

Requirements

• Python 3.8 - 3.12
• PySpark 3.3+

License

MIT License