The abundance of natural fibres can facilitate its cost-effective utilization in fibre-reinforced composites by substituting the synthetic variant. In this study, a vacuum pressurization impregnation (VPI) treatment of semi-matured coir (SMC) fibres is done, followed by a mechano-chemical extraction procedure. After that, tests like thermogravimetric analysis (TGA), x-ray diffraction (XRD) and scanning electron microscopy (SEM) are done to examine the thermal features, crystallinity and morphological aspects of the fibres. The results are compared with non-VPI SMC fibres, extracted through a similar way. It is found that the VPI-treated SMC fibre exhibits higher crystallinity, lower thermal deterioration and reduced porosity as compared to non-VPI SMC fibre. An experimental design is created to systematically assess the breaking strength of these fibres, with predictions made using ANFIS, a machine learning model. Furthermore, particle swarm optimization (PSO) is taken to build a PSO-ANFIS model, for optimizing the ANFIS parameters. The results indicate that the PSO-ANFIS model outperforms standalone ANFIS, yielding superior predictive performance with lower root-mean-square error (RMSE) and higher R2 values, specifically 0.59951 and 0.99563, respectively. These findings support the potential of VPI-treated SMC fibres as effective reinforcements in natural fibre composites and highlight the PSO-ANFIS model’s efficacy for accurate breaking strength prediction. It may be beneficial for similar applications in practical composite material design.