This Travel Time Reliability (TTR) is a crucial aspect of transportation planning and management. It affects individual decisions, scheduling, and productivity, and has significant financial implications for passengers and goods. Traffic congestion is a major factor impacting TTR, which can be classified as recurring (predictable) or non-recurring (unanticipated). Researchers have developed various definitions and measures for TTR and Planning Time Index (PTI) is one of these indexes. Proper communication of TTR is essential, and numerical measures like PTI are commonly used to convey this information to travelers. Machine Learning (ML) models, particularly neural networks, have become increasingly popular for TTR estimation due to their ability to handle complex relationships and high-dimensional data. This study proposes using Fitrnet, a feedforward fully connected neural network, for predicting TTR at a network level. While Fitrnet has been used in other fields of engineering, its application in TTR estimation is novel. The study uses a dataset from the UK government covering the Strategic Road Network from April 2015 to March 2021. Results show that a Fitrnet model with 5 hidden layers can accurately predict PTI, with MAPE values below 10% in most cases, demonstrating the effectiveness of Fitrnet for TTR prediction in smaller datasets. The study contributes to the growing body of research on TTR modeling by proposing a new approach using Fitrnet and applying it to a previously unused dataset.