Systems and methods for enriching modeling tools and infrastructure with semantics
Systems and methods for generating and processing modeling workflows.
1. A method implemented by a modeling system and comprising:
generating metadata associated with a model, the metadata comprising, for each of a plurality of features, a reference distribution of feature values within each of a plurality of data sets used to validate or train the model, wherein the reference distribution of feature values comprises, for each of the feature values, a vector comprising a distribution value indicating a percentage of the plurality of data sets having the feature value;
recording the generated metadata in a structured database; and
detecting unexpected input data during execution of the model, comprising, for each feature of one or more production data sets used by the model:
automatically comparing a production distribution of feature values for the feature with the reference distribution of feature values for the feature recorded in the structured database; and
providing an alert to an external system, when an alert condition is determined to be satisfied based on the comparison, wherein the alert indicates detection of the unexpected input data.
2. The method of claim 1 , wherein the production distribution of feature values is a distribution of feature values within a first subset of the production data sets used by the model and the reference distribution is another distribution of feature values within a second subset of the production data sets used by the model.
3. A modeling system, comprising memory comprising instructions stored thereon and one or more processors coupled to the memory and configured to execute the stored instructions to:
generate metadata associated with a model, the metadata comprising, for each of a plurality of features, a reference distribution of feature values within each of a plurality of data sets used to validate or train the model, wherein the reference distribution of feature values comprises, for each of the feature values, a vector comprising a distribution value indicating a percentage of the plurality of data sets having the feature value;
record the generated metadata in a structured database; and
detect unexpected input data during execution of the model, comprising, for each feature of one or more production data sets used by the model:
automatically compare a production distribution of feature values for the feature with the reference distribution of feature values for the feature recorded in the structured database; and
provide an alert to an external system, when an alert condition is determined to be satisfied based on the comparison, wherein the alert indicates detection of the unexpected input data.
4. A non-transitory computer readable medium having stored thereon instructions comprising executable code that, when executed by one or more processors, causes the one or more processors to:
generate metadata associated with a model, the metadata comprising, for each of a plurality of features, a reference distribution of feature values within each of a plurality of data sets used to validate or train the model, wherein the reference distribution of feature values comprises, for each of the feature values, a vector comprising a distribution value indicating a percentage of the plurality of data sets having the feature value;
record the generated metadata in a structured database; and
detect unexpected input data during execution of the model, comprising, for each feature of one or more production data sets used by the model:
automatically compare a production distribution of feature values for the feature with the reference distribution of feature values for the feature recorded in the structured database; and
provide an alert to an external system, when an alert condition is determined to be satisfied based on the comparison, wherein the alert indicates detection of the unexpected input data.