Predictive Analytics

Build, evaluate, and interpret predictive models — classification, regression, and time series — without writing training loops from scratch.

Highlight functions

CreateBoostedTreeModel

XGBoost model with cross-validation, feature importance plot, and SHAP values for interpretability.

from analysistoolbox.predictive_analytics import CreateBoostedTreeModel

model = CreateBoostedTreeModel(
    dataframe=df,
    outcome_variable="churn",
    list_of_predictor_variables=["tenure", "monthly_charges", "contract_type"],
    model_type="classification"
)

CreateARIMAModel

ARIMA time series forecasting with automatic order selection, residual diagnostics, and forecast plot.

from analysistoolbox.predictive_analytics import CreateARIMAModel

model = CreateARIMAModel(
    dataframe=df,
    date_column="month",
    outcome_variable="revenue",
    number_of_periods_to_forecast=12
)

CreateNeuralNetwork_SingleOutcome

TensorFlow/Keras neural network for single-output classification or regression. Handles training, validation, and loss curve visualization.

All functions

| Function | Description | |---|---| | CreateARIMAModel | ARIMA time series forecasting | | CreateBoostedTreeModel | XGBoost with SHAP interpretability | | CreateDecisionTreeModel | Interpretable decision trees | | CreateExponentialSmoothingModel | Holt-Winters exponential smoothing | | CreateLinearRegressionModel | Linear regression with diagnostics | | CreateLogisticRegressionModel | Logistic regression for binary outcomes | | CreateNeuralNetwork_SingleOutcome | TensorFlow deep learning model |