Transaction classification based on transaction time predictions

1. A method of training a machine learning model, comprising:

obtaining a transaction record associated with a user, the transaction record comprising a first transaction description;

obtaining one or more user positions associated with the user, wherein each user position of the one or more user positions corresponds to a timestamp;

obtaining a plurality of business records, wherein each respective business record of the plurality of business records:

comprises a respective transaction description,

is associated with a user position of the one or more user positions; and

corresponds to an establishment within a threshold range of a user position associated with the respective business record;

comparing the first transaction description to each respective transaction description of the plurality of business records;

calculating a similarity score for each respective business record based on the comparing;

associating the transaction record with a particular business record of the plurality of business records with a highest similarity score;

determining a transaction time for the transaction record based on a timestamp associated with the particular business record;

providing the transaction time for the transaction record as an input to a machine learning model that has been trained based on historical transaction times and historical classifications of historical transactions;

determining, based on an output from the machine learning model in response to the input, a classification for the transaction record;

receiving user input indicating whether the classification is accurate; and

re-training the machine learning model based on the user input and the transaction time for the transaction record.

2. The method of claim 1 , further comprising presenting, in a graphical user interface (GUI), an indication of the classification for the transaction record.

3. The method of claim 1 , wherein each user position of the one or more user positions comprises global positioning system (GPS) coordinates.

4. The method of claim 1 , wherein:

each respective transaction description of the plurality of business records of the one or more business records comprises at least one of a type of a given establishment or a name of the given establishment, and

comparing the first transaction description to each respective transaction description of the plurality of business records comprises comparing at least one of the type of the establishment or the name of the establishment of the respective transaction description with the first transaction description of the transaction record.

5. The method of claim 1 , wherein:

the transaction record further comprises timing data, and

the method further comprises adjusting the timing data of the transaction record using the transaction time.

6. The method of claim 1 , wherein:

the transaction record further comprises timing data, and

the one or more user positions correspond to one or more dates within a search period defined based on the timing data.

7. A non-transitory computer readable medium comprising instructions that, when executed by a processor of a processing system, cause the processing system to perform a method of training a machine learning model, the method comprising:

obtaining a transaction record associated with a user, the transaction record comprising a first transaction description;

obtaining one or more user positions associated with the user, wherein each user position of the one or more user positions corresponds to a timestamp;

obtaining a plurality of business records, wherein each respective business record of the plurality of business records:

comprises a respective transaction description;

is associated with a user position of the one or more user positions; and

corresponds to an establishment within a threshold range of a user position associated with the respective business record;

comparing the first transaction description to each respective transaction description of the plurality of business records;

calculating a similarity score for each respective business record based on the comparing;

associating the transaction record with a particular business record of the plurality of business records with a highest similarity score;

determining a transaction time for the transaction record based on at least a timestamp associated with the particular business record;

providing the transaction time for the transaction record as an input to a machine learning model that has been trained based on historical transaction times and historical classifications of historical transactions;

determining, based on an output from the machine learning model in response to the input, a classification for the transaction record;

receiving user input indicating whether the classification is accurate; and

re-training the machine learning model based on the user input and the transaction time for the transaction record.

8. The non-transitory computer readable medium of claim 7 , wherein the method further comprises presenting, in a graphical user interface (GUI), an indication of the classification for the transaction record.

9. The non-transitory computer readable medium of claim 7 , wherein each user position of the one or more user positions comprises global positioning system (GPS) coordinates.

10. The non-transitory computer readable medium of claim 7 , wherein:

each respective transaction description of the plurality of business records of the one or more business records comprises at least one of a type of a given establishment or a name of the given establishment, and

comparing the first transaction description to each respective transaction description of the plurality of business records comprises comparing at least one of the type of the establishment or the name of the establishment of the respective transaction description with the first transaction description of the transaction record.

11. The non-transitory computer readable medium of claim 7 , wherein:

the transaction record further comprises timing data, and

the method further comprises adjusting the timing data of the transaction record using the transaction time.

12. The non-transitory computer readable medium of claim 7 , wherein:

the transaction record further comprises timing data, and

the one or more user positions correspond to one or more dates within a search period defined based on the timing data.

13. A computer system, comprising:

a memory comprising computer-executable instructions; and

a processor configured to execute the computer-executable instructions and cause the computer system to perform a method of training a machine learning model, the method comprising:

obtaining a transaction record associated with a user, the transaction record comprising a transaction first description;

obtaining one or more user positions associated with the user, wherein each user position of the one or more user positions corresponds to a timestamp;

obtaining a plurality of business records, wherein each respective business record of the plurality of business records:

comprises a respective transaction description;

is associated with a user position of the one or more user positions; and

corresponds to an establishment within a threshold range of a user position associated with the respective business record;

comparing the first transaction description to each respective transaction description of the plurality of business records;

calculating a similarity score for each respective business record based on the comparing;

associating the transaction record with a particular business record of the plurality of business records with a highest similarity score;

determining a transaction time for the transaction record based on a timestamp associated with the particular business record;

providing the transaction time for the transaction record as an input to a machine learning model that has been trained based on historical transaction times and historical classifications of historical transactions;

determining, based on an output from the machine learning model in response to the input, a classification for the transaction record;

receiving user input indicating whether the classification is accurate; and

re-training the machine learning model based on the user input and the transaction time for the transaction record.

14. The computer system of claim 13 , the method further comprising presenting in a graphical user interface (GUI), an indication of the classification for the transaction record.

15. The computer system of claim 13 , wherein each user position of the one or more user positions comprises global positioning system (GPS) coordinates.

16. The computer system of claim 13 , wherein:

each respective transaction description of the plurality of business records of the one or more business records comprises at least one of a type of a given establishment or a name of the given establishment, and

comparing the first transaction description to each respective transaction description of the plurality of business records comprises comparing at least one of the type of the establishment or the name of the establishment of the respective transaction description with the first transaction description of the transaction record.

17. The computer system of claim 13 , wherein:

the transaction record further comprises timing data, and

the method further comprises adjusting the timing data of the transaction record using the transaction time.

18. The computer system of claim 13 , the method further comprising storing a user position associated with the business record associated with the transaction record in a lookup table for assignment to subsequent transaction records with attributes that are similar to those of the transaction record.

19. The method of claim 1 , further comprising storing a user position associated with the business record associated with the transaction record in a lookup table for assignment to subsequent transaction records with attributes that are similar to those of the transaction record.

20. The non-transitory computer readable medium of claim 8 , wherein the method further comprises storing a user position associated with the business record associated with the transaction record in a lookup table for assignment to subsequent transaction records with attributes that are similar to those of the transaction record.