Experts have presented a new model for estimating the cost of housing

Domclick specialists have updated real estate valuation algorithms aimed at improving the accuracy of ML-model calculations, said Alexey Leypi, director of Sberbank's Domclick department, during the Domclick Digital Day conference in Vladivostok.

The changes affected all key areas: the assessment of apartments, suburban real estate and forecasting seasonal changes in rental rates.

Now, to estimate the market value of an apartment, it is enough for users to specify the address and area of the object, no other parameters are required. The ML model will compare the price of an apartment with similar relevant ads on Domclick and real mortgage transactions by Sberbank over the past 3 months. The analysis time will be several minutes.

According to Leypi, the new model has improved the accuracy of the assessment of households — country houses and land plots in Moscow, the Moscow region, St. Petersburg and the Leningrad region. The ML model evaluates suburban real estate throughout Russia, but previously price analysis in the listed regions was difficult due to the high volatility of the market.

The new model will allow landlords to plan the profitability of renting out housing thanks to the function of forecasting the cost of objects for 6 months in advance in the short-term rental service Domklik24. The model will analyze seasonal fluctuations in demand, as well as examine the cost of similar objects in ads and in previously concluded transactions.