Nowcasting abstracted census from individual customs transaction records

1. A method, comprising:

defining, by a computing device, a signal relationship between a granular data value and a target data value, wherein at least a portion of the granular data value corresponds to a granular latency value that is smaller than a target data latency value corresponding to the target data value;

interpreting, by the computing device, granular data corresponding to the granular data value, including predicting one or more missing values in the granular data by:

executing a trained natural language processing (NLP) model using the granular data as input to generate a predicted classification for the granular data,

determining one or more other sets of granular data associated with the predicted classification,

associating the granular data with one of the other sets of granular data when a similarity value between a first mathematical vector corresponding to the granular data and a second mathematical vector corresponding to one of the other sets of granular data is greater than a threshold similarity value, and

determining predicted values for the one or more missing values in the granular data based upon the predicted classification for the granular data and the association between the granular data and the one of the other sets of granular data;

aggregating, by the computing device, the granular data in response to the signal relationship;

estimating, by the computing device, a value of the target data value for a selected time reference by:

applying a time series transformation to the granular data based on the selected time reference, wherein applying the time series transformation comprises performing a differencing operation; and

applying a machine learning model to an output of the time series transformation, wherein applying the machine learning model comprises performing a tree-based method;

determining, by the computing device, a nowcasting prediction of the target data value based on the estimated value of the target data value;

generating, by the computing device, a push notification comprising the nowcasting prediction of the target data value and transmitting the push notification to a mobile device;

activating, by the mobile device, a graphical user interface of the mobile device in response to waking the mobile device from a sleep mode based upon receipt of the push notification; and

displaying, by the graphical user interface of the mobile device, the push notification to a user of the mobile device.

2. The method of claim 1 , wherein aggregating comprises aggregating the granular data value in a hierarchical format.

3. The method of claim 2 , further comprising processing the granular data prior to the aggregating, wherein the processing further comprises indexing the granular data in response to the hierarchical format.

4. The method of claim 2 , further comprising processing the granular data prior to the aggregating, wherein the processing further comprises indexing the granular data in response to the signal relationship.

5. The method of claim 2 , further comprising processing the granular data prior to the aggregating, wherein the processing further comprises verifying the signal relationship in response to at least one sub-aggregation developed during the aggregating.

6. The method of claim 2 , further comprising processing the granular data prior to the aggregating, wherein the processing comprises correcting erroneous data in the granular data.

7. The method of claim 1 wherein applying the time series transformation further comprises performing one or more of a conversion to percent change, convolutional filtering, trend regression, cycle regression, a power transform, and a smoothing operation.

8. The method of claim 1 wherein applying the machine learning model further comprises performing one or more of a linear regression, a regularized regression, applying a support vector machine, applying a neural network, determining a time domain distribution, and determining a convergence in multiple dimensions.

9. The method of claim 1 further comprising identifying a hierarchical relationship between at least two data fields of the granular data.

10. The method of claim 9 wherein identifying the hierarchical relationship further comprises receiving a data structure defining the hierarchical relationship between the at least two data fields of the granular data.

11. The method of claim 1 further comprising weighting a value of the target data value for the selected time reference.

12. The method of claim 11 wherein weighting comprises performing one or more of a linear regression, a regularized regression, applying a support vector machine, a tree-based method, applying a neural network, determining a time domain distribution, and determining a convergence in multiple dimensions.

13. The method of claim 1 further comprising correlating a change in at least one data field of the granular data with a change in the target data value.

14. The method of claim 1 further comprising generating a cached data structure based on an aggregation of the granular data, the cached data structure comprising a plurality of data fields extracted from the aggregation of the granular data.

15. The method of claim 14 further comprising generating an index based on at least two data fields extracted from the aggregation of the granular data.

16. The method of claim 1 further comprising determining feedback data in response to determining the nowcasting prediction of the target data value.

17. The method of claim 16 wherein the feedback data comprises a census report.

18. The method of claim 16 further comprising updating the signal relationship in response to the determined feedback data.

19. The method of claim 1 further comprising providing access to open an application on the user device to access the nowcasting prediction of the target data value.