The following is a brief summary of the methods we use to provide accurate age information.
The data processing is prone to change as the algorithm is trained on user's response to the information provided.
Pulling the year of manufacturer from the serial number is the first step is determining the age of an appliance. The serial number is decoded by running it through the appropriate manufacturer specified convention.
The serial number conventions vary significantly from manufacturer to manufacturer, as well as the amount of data encoded into it. Some manufacturers will encode the exact year down to the month while others will encode the last digit of the year alone.
A comprehensive list of serial number date codes for major appliance brands can be found here.
For major appliances like HVAC and water heaters, serial conventions can be found here and here respectively.
A significant portion of our serial number decoding logic is derived from those resources.
More often than not, a serial number will provide multiple possible years of manufacture.
This occurs when a manufacturer does not encode precise manufacturing data into the serial number or the date codes in the serial number are recycled (which occurs about every 20-30 years on average).
When this happens, we use the model number to narrow down which of the dates found in the serial number is the most likely.
Each data point we can find on the model is compared to the possible dates that were decoded from the serial number. These data points are weighted differently depending on the source of the data.
The weight of the data point fluctuates as we receive feedback on decode results. The below information is a general overview.
Extracting dates from product support pages, owners manuals, service manuals and more directly from the manufacturer.
Finding, filtering and parsing public reviews of the model to glean date information.
Crawling listings of the model to determine if it is still being sold or if it has been discontinued.
Searching our internal databases for positive feedback on other decode results of the model.
Private access to many third party databases.
Performing a sweep of the internet for mentions of the model. This data is generally not reliable and has the lowest weight of any data point.
Each data point found is then iteratively compared to the possible dates we retrieved from the serial number.
We then come up with a confidence score based on the source of the data points we could find and the serial number year options.
It is never an exact science and a significant amount of inference and assumption must be made. Due to this we only allow our confidence algorithm a max of 80% since there is always room for error.
It is unreasonably difficult to accurately determine the age of an appliance reliably.
Many factors contribute to the difficulty such as incorrect user inputs, lack of documentation from the manufacturer, model number differences between retailers and most importantly; the constantly changing serial number date code conventions that select manufacturers must update annually.
We hope the information provided by us is accurate and helpful. If you have any questions or concerns, feel free to contact us.
Copyright homespy.io 2024