Electrospun structures, known for their high porosity and surface area, can be tuned by optimising manufacturing parameters. These characteristics make them ideal for waterproof and breathable textiles, skin-like non-woven fabrics, and smart wearable bioelectronic textiles. This research aims to develop a manufacturing optimisation methodology using machine learning models to control fibre diameter and inter-fibre separation for textile applications. Polyvinyl alcohol (PVA) structures were produced with varying concentrations (10, 12, 14, 16 w/v) and different parameters such as flow rate (0.5–5 ml/h), voltage (18–25 kV), needle diameter (15–23 G), distance between needle and collector (5–11 cm), and mandrel revolution (500–3000 rpm). Data from 2560 observations of fibre diameter and inter-fibre separations were used to train 20 machine learning models. C5.0 Decision Trees and Rule-Based Models identified optimal setups, achieving high prediction accuracy for fibre diameter (0.868) and inter-fibre separation (0.861). This research advances the optimisation of electrospinning techniques for textile applications.