Systems and methods for networks of time series

Abstract

This patent specification relates to systems and methods that use networks of standardized time series and models in a generic and extensible platform. More particularly, this patent specification relates to standardizing time series and models, using standardized time series and models in a network of time series (NOTS), and creating networks of time series in a NOTS platform. In addition, this patent specification relates to use of nodes that can be arranged in any user defined order, and each node is evaluated to return a time array that may be used to populate a time series. A layering framework may be used to define how each node is evaluated.

1 . A computer implemented method for using a network of time series platform comprising:
 defining a network of time series (NOTS) comprising a combination of time series arranged in an upstream to downstream format in which any upstream time series directly or indirectly feeds into a downstream time series, wherein defining the NOTS comprises:
 receiving user input via the NOTS platform selection of each time series for inclusion into the combination of time series, where each time series is selected from one of a source time series, a transform time series, and a model time series; 
 arranging, via user input on the NOTS platform, a position of each time series comprising the combination of time series within the upstream to downstream format of the NOTS; and 
 wherein the combination of time series comprises a plurality of source time series and at least one model time series; 
   generating predicted values for the at least one model time series using the NOTS; and   performing an action based on the generated predicted values.   
     
     
         2 . The method of  claim 1 , wherein the plurality of time series comprises exogenous time series and endogenous time series, wherein a first model of the plurality of models is fit with at least one first exogenous time series and one first endogenous time series, wherein the first model is operative to generate predicted values based on the at least one first exogenous time series, wherein values associated with the at least one first endogenous time series are provided to a first model time series of the plurality of model time series for all combinations of natural time intervals and as of time intervals, and wherein the predicted values for the at least one first endogenous time series generated by the first model are provided to the first model time series and used to overwrite the values provided by the at least one first endogenous time series for specific combinations of natural time intervals and as of time intervals. 
     
     
         3 . The method of  claim 1 , wherein each of the source time series, transform time series, and the model time series comprises a natural time axis and an as of time axis. 
     
     
         4 . The method of  claim 1 , wherein a source time series, transform time series, or a model time series within the NOTS is characterized as one of an upstream time series, a downstream time series, and both an upstream time series and a downstream time series relative to any other source time series, transform time series, or model time series within the NOTS. 
     
     
         5 . The method of  claim 4 , wherein the downstream time series may have a higher intrinsic value than an intrinsic value of upstream time series, and wherein downstream time series produce values that may have lower confidence or a lower accuracy than values produced by upstream time series. 
     
     
         6 . The method of  claim 1 , wherein at least one of the plurality of source time series comprises future predicted values. 
     
     
         7 . The method of  claim 1 , wherein at least one of the model time series is used as an input to one of the plurality of models. 
     
     
         8 . The method of  claim 1 , wherein the generated predicted values are future predicted values. 
     
     
         9 . The method of  claim 1 , wherein the generated predicted values are past predicted values. 
     
     
         10 . The method of  claim 1 , wherein the generated predicted values are present predicted values. 
     
     
         11 . The method of  claim 1 , wherein the generated predicted values are relative to an as of time. 
     
     
         12 . The method of  claim 1 , wherein the generated predicted values are not relative to an as of time. 
     
     
         13 . The method of  claim 1 , wherein the action is performed external to the network of time series platform. 
     
     
         14 . A computer implemented method for using time series, comprising:
 receiving user inputs via a platform to define a network of time series;   receiving user inputs via the platform to define an evaluation time period; and   evaluating the network of time series based on the evaluation time period, comprising:
 accessing at least one exogenous time series that supplies values to a model; 
 accessing at least one endogenous time series, wherein the model is configured to predict values associated with the endogenous time series, and wherein the endogenous time series serves as an underlying time series for a model time series; 
 populating the model time series with values supplied by the underlying endogenous time series for all combinations of natural time intervals and as of time intervals; 
 using the at least one exogenous time series and the at least one endogenous time series to fit the model; 
 using the fitted model to generate predicted values based on the at least one exogenous time series; 
 using the predicted values to overwrite at least one of the underlying values in the model time series for specific combinations of natural time intervals and as of time intervals; and 
 using the model time series as an input to a second model or as a basis for a user to perform an action. 
   
     
     
         15 . The method of  claim 14 , wherein the model time series comprises a natural time axis and an as of time axis, wherein an as of time interval within the as of time axis represents a time from which a predicted value is projected across the natural time axis. 
     
     
         16 . The method of  claim 15 , wherein the predicted values for a specific as of time interval are relative to the as of time interval. 
     
     
         17 . The method of  claim 16 , wherein the model time series comprises the predicted values for natural time intervals relative to the as of time interval, wherein a range of natural time intervals spans from a start offset through an end offset. 
     
     
         18 . The method of  claim 17 , wherein the range of natural time intervals exists entirely before the as of time interval, spans across the as of time interval, or exists entirely after the as of time interval. 
     
     
         19 . The method of  claim 15 , wherein the predicted values for a specific as of time interval are not relative to the as of time interval. 
     
     
         20 . The method of  claim 19 , wherein the model time series comprises the predicted values for a fixed range of natural time intervals, wherein the range of natural time intervals spans from a start time to an end time. 
     
     
         21 . The method of  claim 14 , wherein the at least one exogenous time series comprises known values and predicted future values. 
     
     
         22 .- 93 . (canceled)