We’ve interviewed our Data Scientists Qiang Huang and Diem Nguyen to study how SkenarioLabs forecasts the value and potential of properties using the AVM model and how businesses can benefit from it.
1. Can you please explain briefly how SkenarioLabs forecasts the value and potential of properties using the AVM model?
SkenarioLabs AVMs are built using the combination of geospatial and economic data, physical building attributes and machine learning. To create accurate and reliable AVMs, we use the market approach and methodology described in International Valuation Standards 2017. The algorithms we use capture complex nonlinear interactions between variables and thereby we achieve a high degree of granularity. Due to the fact that we utilize regional demographic and economic data our model is not dependent on the similar transactions carried out nearby. Therefore our model performs in areas with low market liquidity. In addition to market prices, we also provide a price corrected with possible mispricing of the future renovations and value forecasts for the next 5 years.
2. How can Financial Institutes benefit from SkenarioLabs service?
In credit decision purpose, financial institutes now can use AVMs as an alternative option to check whether the proposed value of a given collateral asset is acceptable with low cost first, before making further decisions and spending more money to order a full valuation by a human appraisal. In addition, AVMs help increase transparency, manage risks, and enhance loan-level data.
3. Are there any other types of industries that would benefit from SkenarioLabs’ AVM model? If yes, which ones and why?
Yes absolutely! In audit domain, lenders can use AVMs as secondary resources to double-check whether values derived from human values are accurate.
In fraudulent detection domain, AVMs can evaluate many properties prices at a time and hence detect those do not follow the normal market trend. In public authorities, AVMs help calculating compensation payment for homeowners as expenses of new public work (i.e. road, highway, airport expansion, etc.). Besides, investors could also leverage the use of AVMs in estimating expected tax capital for either single or portfolio properties for tax planning purposes. Last but not least, the real estate manager can leverage our AVMs to manage the performance of every property in their portfolio. In greater detail, they can gain more insights about property incomes versus expenses, yield and internal rate of return, which is very beneficial for their decision-making.
4. What are your biggest challenges?
One of the biggest challenges is that the rural area has very few transactions. The reliability of data is another concern, as there are wrong prices in transaction dataset. Last but not least, the accuracy of the property attributes is not accurate, for example, the addresses and house attributes that customers provide are often incorrect. Therefore, we make sure that all of these aspects are fixed on a daily basis.
5. What are the upcoming product releases?
In the future, we’ll use deep learning and map images to measure the characteristics of a neighbourhood. We will add more distance features inside the same neighbourhood and more will be shortly announced! Stay tuned!
The interview is now over, but if you would like to know more about this topic, don’t hesitate to contact our experts in data Diem at firstname.lastname@example.org or Qiang at email@example.com. You can also send us a direct message by filling your information below.