Using transaction data to predict vehicle depreciation and present value

Abstract

Various embodiments are directed to a system or platform with machine learning capabilities configured to accurately predict in real-time a depreciation factor of a vehicle associated with a customer and further accurately predict a present value of the vehicle based at least in part on card transaction data associated with the customer. Based on one or more factors, such as a determination that the present value of the vehicle falls below a predefined threshold value, one or more auto financing products may be generated and provided to the customer by the system.

What is claimed is: 
     
         1 . A system comprising:
 at least one computing device comprising:   memory to store instructions; and   one or more processors, coupled with the memory, operable to execute the instructions that, when executed, cause the one or more processors to:   receive or access customer transaction data associated with a card belonging to a customer, the customer transaction data including a plurality of expenses;   determine which of the plurality of expenses are fuel expenses based on an analysis of the customer transaction data;   determine at least a make, model, and year of a vehicle associated with the customer;   determine fuel consumption information for the make, model, and year of the vehicle;   determine a cost corresponding to each determined fuel expense;   determine a predicted number of miles driven on the vehicle based at least in part on: (i) the fuel expenses, (ii) the cost corresponding to each fuel expense, and (iii) the fuel consumption information for the vehicle;   determine, via a machine learning model, a depreciation factor associated with the vehicle based on the predicted number of miles driven on the vehicle; and   determine, via the machine learning model, a present value of the vehicle based at least in part on the determined depreciation factor.   
     
     
         2 . The system of  claim 1 , wherein the one or more processors is further caused to:
 determine whether the present value of the vehicle falls below a predefined threshold value;   generate one or more auto financing products based on the determination that the present value falls below the predefined threshold value; and   provide the one or more auto financing products to the customer via one or more channels.   
     
     
         3 . The system of  claim 1 , wherein the fuel consumption information comprises a city miles per gallon (MPG) value associated with a city environment and a highway MPG value associated with a highway environment and wherein the one or more processors is further caused to average the city MPG value and highway MPG value to obtain an average MPG associated with the vehicle. 
     
     
         4 . The system of  claim 1 , wherein the determination of which of the plurality of expenses in the customer transaction data are fuel expenses comprises the one or more processors being further caused to analyze each of the plurality of expenses based on a merchant category code (MCC) and identify an expense as a fuel expense if the analyzed MCC corresponds to a fuel station MCC or a fuel-related MCC. 
     
     
         5 . The system of  claim 1 , wherein the determination of which of the plurality of expenses in the customer transaction data are fuel expenses comprises the one or more processors being further caused to analyze each of the plurality of expenses based on textual descriptors and identify an expense as a fuel expense if the analyzed textual descriptors indicate that the fuel expense is related to a fuel station or fuel. 
     
     
         6 . The system of  claim 1 , wherein the determination of which of the plurality of expenses in the customer transaction data are fuel expenses comprises the one or more processors being further caused to analyze a pattern or a trend in the plurality of expenses and identify an expense as a fuel expense based on the analyzed pattern or the analyzed trend in the plurality of expenses. 
     
     
         7 . The system of  claim 1 , wherein the plurality of expenses is derived from a monthly card statement associated with the card and the predicted number of miles driven corresponds to a time period associated with the monthly card statement. 
     
     
         8 . The system of  claim 7 , wherein the one or more processors is further caused to determine a total number of miles driven on the vehicle in a year by projecting out the predicted number of miles driven. 
     
     
         9 . The system of  claim 1 , wherein the machine learning model is a classification model, a linear regression model, a logistic regression model, a decision tree model, a support vector machine (SVM) model, a Naïve Bayes model, a k-nearest neighbors (kNN) model, k-means model, a random forest model, a dimensionality reduction algorithm, a gradient boosting algorithm, an XGBoost algorithm, a LightGBM algorithm, or a Catboost model. 
     
     
         10 . The system of  claim 1 , wherein the one or more processors is further caused to receive real-world data related to the vehicle and cause the machine learning model to learn or adjust the depreciation factor of the vehicle based on the real-world data. 
     
     
         11 . A method comprising:
 receiving or accessing customer transaction data associated with a card belonging to a customer, the customer transaction data including a plurality of expenses;   determining, via one or more processors, which of the plurality of expenses are fuel expenses based on an analysis of the customer transaction data;   determining at least a make, model, and year of a vehicle associated with the customer;   determining fuel consumption information for the make, model, and year of the vehicle;   determining a cost corresponding to each fuel expense;   determining, via the one or more processors, a predicted number of miles driven on the vehicle based at least in part on: (i) the fuel expenses, (ii) the cost corresponding to each fuel expense, and (iii) the fuel consumption information for the vehicle; and   determining, via a machine learning model, a depreciation factor associated with the vehicle based on the predicted number of miles driven on the vehicle.   
     
     
         12 . The method of  claim 11 , further comprising:
 determining, via the machine learning model, a present value of the vehicle based at least in part on the determined depreciation factor.   
     
     
         13 . The method of  claim 12 , further comprising:
 determining, via the one or more processors, whether the present value of the vehicle falls below a predefined threshold value;   generating one or more auto financing products based on the determination that the present value falls below the predefined threshold value; and   providing the one or more auto financing products to the customer via one or more channels.   
     
     
         14 . The method of  claim 11 , wherein the customer is both a card customer and an auto financing customer. 
     
     
         15 . The method of  claim 11 , further comprising:
 receiving real-world data related to the vehicle; and   causing the machine learning model to learn or adjust the depreciation factor of the vehicle based on the real-world data.   
     
     
         16 . A non-transitory computer-readable storage medium storing computer-readable program code executable by at least one processor to:
 receive or access customer transaction data associated with a card belonging to a customer, the customer transaction data including a plurality of expenses;   determine which of the plurality of expenses are fuel expenses based on an analysis of the customer transaction data;   determine at least a make, model, and year of a vehicle associated with the customer;   determine fuel consumption information for the make, model, and year of the vehicle;   determine a cost corresponding to each determined fuel expense;   determine a predicted number of miles driven on the vehicle based at least in part on: (i) the fuel expenses, (ii) the cost corresponding to each fuel expense, and (iii) the fuel consumption information for the vehicle; and   determine, via a machine learning model, a depreciation factor associated with the vehicle based on the predicted number of miles driven on the vehicle.   
     
     
         17 . The non-transitory computer-readable storage medium of  claim 16 , wherein the computer-readable program code further causes the at least one processor to determine, via the machine learning model, a present value of the vehicle based at least in part on the determined depreciation factor. 
     
     
         18 . The non-transitory computer-readable storage medium of  claim 17 , wherein the computer-readable program code further causes the at least one processor to:
 determine whether the present value of the vehicle falls below a predefined threshold value;   generate one or more auto financing products based on the determination that the present value falls below the predefined threshold value; and   provide the one or more auto financing products to the customer via one or more channels.   
     
     
         19 . The non-transitory computer-readable storage medium of  claim 16 , wherein the machine learning model is a classification model, a linear regression model, a logistic regression model, a decision tree model, a support vector machine (SVM) model, a Naïve Bayes model, a k-nearest neighbors (kNN) model, k-means model, a random forest model, a dimensionality reduction algorithm, a gradient boosting algorithm, an XGBoost algorithm, a LightGBM algorithm, or a Catboost model. 
     
     
         20 . The non-transitory computer-readable storage medium of  claim 16 , wherein the computer-readable program code further causes the at least one processor to:
 receive real-world data related to the vehicle; and   cause the machine learning model to learn or adjust the depreciation factor of the vehicle based on the real-world data.