I have a list of incomplete (e.g. missing the city field) or incorrect (e.g. spelling mistakes) human-readable addresses to which my company car needs to go in the future. So for one wrong address, there could be different possible address guesses that are 1000 km away from each other. So for now it's impossible for me to geocode the addresses into coordinates (longitude and latitude).

What I have is a list of GPS coordinates that my car returned for the trips in the past. Every time the company car goes on a trip, it returns its GPS locations every 10 minutes. Also there are only a certain number of regions and cities for my company car to go, so let's assume almost all of the incomplete or incorrect addresses are in the same cities that my car went before.

I was suggested to use The Place Autocomplete service to guess the addresses. But since I have the historical data, I was wondering there is an algorithm for the guessing.

Thank you.

CodePudding user response：

There exists many algorithms which compare two strings (e.g. two addresses) and output a similarity rate. For instance : Predict function of T-SQL.