With Additive Manufacturing (AM) technology such as Fused Deposition Modeling (FDM) technique being used to produce functional components, it is necessary to understand the combined effect of significant printing process parameters over the mechanical properties of printed parts. The present work employed the full factorial technique to investigate the effect of printing process parameters such as infill density, infill pattern, layer height, and nozzle size over the tensile strength of the printed parts. Analysis of Variance (ANOVA) was conducted and it identified the nozzle size as the most significant factor influencing tensile strength, followed by infill density. Layer height and infill shape had a smaller individual impact on tensile strength. However, with specific combinations of nozzle size and infill densities, noticeable variations in tensile strength were observed. Increasing the infill density enhances tensile strength proportional to the rise in mass due to the additional material. Although infill patterns had minimal effect on tensile strength, the specific strength varied, with the triangle pattern showing the highest specific strength of 7.11 MPa/g, which is 5.20% and 21.65% higher than the rectilinear and wiggle patterns. The highest tensile strength of 43.63 MPa was achieved using the wiggle pattern at 80% infill which is due to print orientation of wiggle pattern with the tensile load. Further, increasing the layer height and nozzle size significantly improved specific strength because of higher print quality and reduced defects. The experimental investigation proved the optimal nozzle to layer height ratio (N/L) for to achieve greater strength is 1.66. With the extensive datasets obtained using experimental investigation, a machine learning model was trained for predicting the tensile strength for the given printing process parameters. Due to its adaptability, efficiency and robustness, the Gaussian Process Regression was proved to estimate the tensile strength of the Polylactic Acid (PLA) material with more accuracy. The predictive performance and corresponding residuals of the training and testing datasets resulted with MAE of 3.17, MAPE of 11.66, and an accuracy of 88.34%.