I'm wondering how Application Insights infers the client's device model
because the Microsoft docs do not state what method is used. Obviously, they must pull it from the client's user agent string, but do they use some sophisticated ML algorithm to classify the device model? Or do they simply apply some regex based logic?
I'm asking because I'm not sure how reliable this information is and I'm considering to use it as normalized input for an own multi-class classifier to categorize the user agents into four classes (mobile, desktop, tablet, unknown).