Implementation

The demand forecast for long-distance trains is created both for 90 days - the "depth" of ticket sales in railway transport, and for an entire year. The predictive analytics system developed by the Tsifromatika’s team provides a forecast for a variety of parameters: by route section, by individual train, by car type, and even by upper and lower berths. Based on this forecast, as well as taking into account a number of parameters (for example, restrictions on the train length), the algorithm generates the optimal scheme of the train.

The solution is based on big data technologies. Forecasting models are built on the basis of railway ticket sales data for the last three years (including additional parameters - competitor prices, holidays, promotions). At the "depth" of the sale, the forecast is recalculated every day, taking into account new incoming information.

The solution works as a separate independent system. At the same time, it is integrated with data sources - the Russian Railways database on train schedules and their belonging, the tariff setting system, the database of competitors' tariffs (air and bus transportation), with the car fleet management system. Thus, the predictive analytics system is at the same time FPC's own consolidated transportation database.

The technical integration of the system with the existing systems of Russian Railways was preceded by extensive work in coordination with various departments of Russian Railways - from the IT department to the security department.