xgbexcel

Python package that converts an XGBRegressor or XGBClassifier models to an Excel formula expression.

NB: The package works with xgboost==1.7.1. In xgboost==2.0, vector leaves were introduced. While I may have limited time to address this myself, contributions and improvements are always welcome. Start from here

How to start

First, you have to install the package.

pip install xgbexcel

How to convert XGBRegressor model to an Excel formula (same for XGBClassifier)

1. Load packages

from xgbexcel import XGBtoExcel
import numpy as np
from xgboost import XGBRegressor

2. Create dummy dataset and fit XGBRegressor model

X_train, y_train = np.random.randint(0, 1000, (100, 2)), np.random.randint(0, 10, 100)
model = XGBRegressor(n_estimators=2, max_depth=1)
model.fit(X_train, y_train)

3. Convert XGBRegressor model to an Excel formula

xgb_excel_expr = XGBtoExcel(model)
xgb_excel_expr.expression

4. The features in the Excel formula are represented using the x1, x2, x3, etc. notation, where the numbers correspond to the enumeration of the features in the XGBRegressor model and X_train. You can manually rename the features in the Excel formula to match the desired column names in the Excel sheet. Once you have renamed the features, you can copy the expression in the Excel sheet as a formula.

feature_map = {'x1': 'feature1', 'x2': 'feature2'}
xgb_excel_expr.rename_features(feature_map)
xgb_excel_expr.expression

5. Save Excel expression to a file

xgb_excel_expr.save_expr('dummy.txt')

6. Enjoy

Try it yourself in the example notebook: howXGBtoExcel.ipynb