Wire Arc Additive Manufacturing (WAAM) is an emerging process for the layer-by-layer production of large, complex metal components. An important aspect is that the layer geometry is influenced by the welding parameters. In this study, the influence of welding parameters, in particular robot travel speed and welding current, on the geometrical properties of the weld beads, including height, width and cross-sectional area, is investigated. A full factorial design of experiments matrix was used to collect data from 50 weld lines. Geometric features were extracted from the point cloud data of the weld lines obtained by laser scanning and analysed using machine learning models. The response surface method was used to model and visualise the relationships. The results show that the welding current has a significant influence on the welding geometry, while the travel speed has a smaller influence. The models showed better accuracy for the height and width of the weld (R2 ≈ 0.87) but were less accurate for the cross-sectional area (R2 ≈ 0.85). Future improvements are indicated, including higher order polynomial terms and additional interaction effects to increase model accuracy.