How much each feature contributes to the XGBoost model's predictions, measured by mean absolute SHAP value and normalised to a share of the total.
Two features carry the model. Bedrooms (35.7%) and neighbourhood (34.3%) are statistically co-dominant, together accounting for 70% of the model's predictive power. Furnishing, payment terms, property type, amenities, recency and engagement make up the remaining 30% combined. This corrects the Issue 01 hunch that location alone sets the price: where a home is and how big it is matter about equally.