The dataset provides the number of patrons that boarded and alighted a particular train service per business date. Estimated passenger counts includes all persons aged 5 and over, excluding drivers and station staff. Data users should interpret the data cautiously, as the model provides estimates only, and its algorithms rely on a series of assumptions that are listed in the Data Quality section below. In particular, greatest caution is needed in relying on estimated train loadings at busier stations where there are greater transfers between train services, such as at the City Loop station and other transfer hubs. Methodologies for the estimation of patronage are subject to continuing improvements as new technology and techniques are available, and this data may be refined and updated over time. Data Quality 1) Passenger counts is derived from a model which includes source data of manual survey counts, myki ticketing data, myki barrier data and manual conductor counts. The purpose of the patronage survey is to determine the transaction rate, which is the percentage of passengers who ‘touch-on’ when they travel. Ticketing transactions are boosted according to the transaction rate to provide an estimate of total patronage. 2) As of January 2021, metropolitan train patronage uses barrier count data, where available, in place of survey observations to determine the transaction rate. 3) Metropolitan train patronage totals are derived from a count of station entries, plus a 5% ‘transfer uplift’ to account for trips following a transfer within a station. Those reported figures will not reconcile with the totals shown here due to the differing methods and application of rounding. 4) Patronage data for regional train is provided by V/Line and is based on conductor or driver counts for each regional train and coach service. 5) Business Date is used instead of the actual date for all fields. The Day of Week and Day Type will follow the Business Date and not the calendar day. 6) Due to the granular nature of the data, the fields of ‘Passenger Boardings’, ‘Passenger Alightings’, ‘Passenger Arrival Load’, and ‘Passenger Departure Load’ have been aggregated into Bin sizes to protect passenger privacy. 7) For Direction, ‘U’ (up) denotes services travelling towards Flinders Street Station and ‘D’ (Down) denotes services travelling away from Flinders Street Station. 8) Line and Group is not differentiated for V/Line services and will have identical cells for V/Line services. 9) For time intervals, 0 is defined as 00:00 (minutes:seconds) to 29:59 and 30 is defined as 30:00 to 59:59 10) Non-timetabled services that ran within the paper ticket section of the train system are not included. This affects passenger counts for V/Line services at non-Myki stations. 11) Services lacking a patronage data source are recorded as having zero counts. It should be noted that in actuality, the passenger counts are unknown and are unlikely to be zero. 12) It is difficult to infer and validate the correct service a patron transferred to in locations where there are multiple transfer options, so the data’s accuracy around these services is reduced. 13) The model is not capacity-constrained, leading to some few services with unrealistically high service load. Such services should be assumed to be very crowded, but these reported loads should not be treated as fact. 14) Where atypical events occurred (e.g. disruptions), the accuracy of the data is significantly reduced. 15) Some trips between Southern Cross and Flinders Street contain ‘dummy’ in their train number field. Due to model limitations, the exact service caught by these passengers is not known, although there is reasonable confidence that these passengers travelled between Southern Cross and Flinders Street at approximately the time shown.