For the Moscow housing market is from a few thousand to tens of thousands of apartments, depending on the selected databases and calculation step (weekly, monthly, quarterly). This is the number of equations is multiplied by another number of periods (eg weeks) during which the model is considered. In other words, the formulated problem is quite complicated and almost not be solved exactly. Therefore, its solution requires the use of various approximations. There are two main approaches to the construction of approximations. The first is the construction of approximate dependency Ak (pi) and Bk (pi) for each individual parameter on the basis of statistics for the period. Such as the function decline sales price of the distance to the subway or the value function of the kitchen area, etc.
As a result of knowing these functions and a set of prices Apartments for Ck (t, pi) at a given time can be calculated set of Gk (t), for a given time. Ideally, all Gk (t), calculated in this way must match. In practice, the observed scatter of these values for the different apartments, which is a consequence of the approximation. If the resulting spread is not very large, it serves as an indicator of adequacy of selection functions Ak (pi) and Bk (pi). As a result, the value of the function G (t) – cost index for the current date – we can calculate, for example, by averaging: G (t) = k (t)> where the symbols denote the averaging operation over all flats of the current period.