valuations general description

• This function gives the user access to home valuations from our Automated Valuation Model (AVM)
• The user sends in home category, coordinates and other input variables and gets a valuation back
• This function is fast and can handle large volumes
• Accuracy ranges from excellent, in the case of inner-city apartments, to low, in the case of detached homes in sparsely populated inland areas in the north
• We also prepare valuations on our residential registry that can be fetched in bulk: https://github.com/Valueguard-Index-Sweden/valueguard-python-client/wiki/residential-registry-valuations

Deviation across different quality levels

In December of 2020, Valueguard analyzed uncertainty levels for our valuations. Valuations were made with all parameters, i.e. no missing fields such as build year. This is the result.

Quality	Median absolute deviation	Share within 10%	Deviation at 95th percentile
	Detached house - Flat	Detached house - Flat	Detached house - Flat
1 - high uncertainty	46% - 47%	0% - 0%	586% - 47%
2 - uncertain	26% - 21%	22% - 25%	112% - 55%
3 - decent	17% - 10%	30% - 48%	70% - 38%
4 - good	9% - 6%	52% - 71%	34% - 21%
5 - high certainty	6% - 6%	76% - 79%	19% - 18%

Share of valuations in each category, per feb-mar 2022:

Quality	Share
1	<0.5%
2	8%
3	29%
4	61%
5	3%