Systems and methods for indicator identification
One aspect of a method for data processing includes identifying target time series data for a target metric and candidate time series data for a plurality of indicators predictive of the target metric; training a machine learning model to predict the target time series data based on the candidate time series data; computing first through third predictivity values based on the machine learning model, wherein the first predictivity value indicates that a source indicator from the plurality of indicators is predictive of the target metric, the second predictivity value indicates that an intermediate indicator from the plurality of indicators is predictive of the target metric, and the third predictivity value indicates that the source indicator is predictive of the intermediate indicator; and displaying a portion of the candidate time series data corresponding to the intermediate indicator and the source indicator based on the first through third predictivity values.
1 . A method for data processing, comprising:
identifying, by a data component, target time series data for a target metric and candidate time series data for a plurality of indicators predictive of the target metric, wherein each of the plurality of indicators comprises a metric different from the target metric;
training, by a training component, a machine learning model to predict the target time series data based on the candidate time series data;
identifying, by a prediction component, a plurality of indicator pairs including a first pair comprising a source indicator from the plurality of indicators and the target metric, a second pair comprising an intermediate indicator from the plurality of indicators and the target metric, and a third pair comprising the source indicator and the intermediate indicator;
computing, by the prediction component, a first predictivity value for the first pair, a second predictivity value for the second pair, and a third predictivity value for the third pair based on the machine learning model and the plurality of indicator pairs, wherein the first predictivity value indicates that the source indicator from the plurality of indicators is predictive of the target metric, wherein the second predictivity value indicates that the intermediate indicator from the plurality of indicators is predictive of the target metric, and wherein the third predictivity value indicates that the source indicator is predictive of the intermediate indicator; and
generating, by a graph component, a graph including the target metric and the plurality of indicators based on the first predictivity value, the second predictivity value, and the third predictivity value, wherein the graph includes a first edge between the source indicator and the target metric, a second edge between the intermediate indicator and the target metric, and a third edge between the source indicator and the intermediate indicator.
2 . The method of claim 1 , further comprising:
displaying, by a user interface, a portion of the candidate time series data corresponding to the intermediate indicator and the source indicator based on the first predictivity value, the second predictivity value, and the third predictivity value.
3 . The method of claim 1 , further comprising:
collecting, by the data component, additional time series data for the intermediate indicator and the source indicator based on the first predictivity value, the second predictivity value, and the third predictivity value; and
generating, by the prediction component, a prediction for the target metric based on the additional time series data.
4 . The method of claim 1 , further comprising:
identifying, by the prediction component, a first time horizon, wherein the source indicator is predictive of the target metric at the first time horizon;
identifying, by the prediction component, a second time horizon that is less than the first time horizon, wherein the intermediate indicator is predictive of the target metric at the second time horizon; and
identifying, by the prediction component, a third time horizon based on the first time horizon and the second time horizon, wherein the source indicator is predictive of the intermediate indicator at the third time horizon.
5 . The method of claim 4 , wherein:
the third time horizon is equal to a difference between the first time horizon and the second time horizon.
6 . The method of claim 4 , wherein:
the first time horizon, the second time horizon, and the third time horizon are identified based on the machine learning model.
7 . The method of claim 6 , further comprising:
identifying, by the prediction component, prediction error values for each of a plurality of time horizons, wherein the first time horizon, the second time horizon, and the third time horizon are identified based on the prediction error values.
8 . The method of claim 4 , further comprising:
receiving, by the prediction component, time horizon input from a user, wherein the first time horizon, the second time horizon, and the third time horizon are identified based on the time horizon input.
9 . The method of claim 1 , further comprising:
generating, by the prediction component, predictivity values for the plurality of indicators, respectively, based on the machine learning model; and
ranking, by the prediction component, the plurality of indicators based on the predictivity values.
10 . The method of claim 1 , further comprising:
ranking, by the prediction component, the plurality of indicator pairs.
11 . A method for data processing, comprising:
identifying, by a data component, target time series data for a target metric and candidate time series data for a plurality of indicators predictive of the target metric;
training, by a training component, a machine learning model to predict the target time series data based on the candidate time series data;
identifying, by a prediction component, a plurality of indicator pairs including a first pair comprising a source indicator from the plurality of indicators and the target metric, a second pair comprising an intermediate indicator from the plurality of indicators and the target metric, and a third pair comprising the source indicator and the intermediate indicator;
identifying, by the prediction component, a first time horizon for the source indicator from the plurality of indicators, wherein the source indicator is predictive of the target metric at the first time horizon;
identifying, by the prediction component, a second time horizon for the intermediate indicator from the plurality of indicators, wherein the intermediate indicator is predictive of the target metric at the second time horizon;
identifying, by the prediction component, a third time horizon for the source indicator, wherein the source indicator is predictive of the intermediate indicator at the third time horizon; and
generating, by a graph component, a graph including the target metric and the plurality of indicators based on the first time horizon, the second time horizon, and the third time horizon, wherein the graph includes a first edge between the source indicator and the target metric, a second edge between the intermediate indicator and the target metric, and a third edge between the source indicator and the intermediate indicator.
12 . The method of claim 11 , wherein:
the second time horizon is less than the first time horizon; and
the third time horizon is based on the first time horizon and the second time horizon.
13 . The method of claim 12 , wherein:
the third time horizon is equal to a difference between the first time horizon and the second time horizon.
14 . The method of claim 11 , wherein:
the first time horizon, the second time horizon, and the third time horizon are identified based on the machine learning model.
15 . The method of claim 14 , further comprising:
identifying, by the prediction component, prediction error values for each of a plurality of time horizons, wherein the first time horizon, the second time horizon, and the third time horizon are identified based on the prediction error values.
16 . The method of claim 11 , further comprising:
receiving, by the prediction component, time horizon input from a user, wherein the first time horizon, the second time horizon, and the third time horizon are identified based on the time horizon input.
17 . The method of claim 16 , further comprising:
generating, by the prediction component, predictivity values for the plurality of indicators, respectively, based on the machine learning model; and
ranking, by the prediction component, the plurality of indicators based on the predictivity values.
18 . The method of claim 11 , further comprising:
identifying, by the prediction component, a plurality of indicator pairs, wherein each of the plurality of indicator pairs includes a first element and a second element; and
ranking, by the prediction component, the plurality of indicator pairs, wherein the source indicator and the intermediate indicator are identified based on the ranking.
19 . An apparatus for data processing, comprising:
a processor;
a memory storing instructions executable by the processor;
a data component executable by the processor to identify target time series data for a target metric and candidate time series data for a plurality of indicators predictive of the target metric;
a training component executable by the processor to train a machine learning model to predict the target time series data based on the candidate time series data;
a prediction component executable by the processor to;
identify a plurality of indicator pairs including a first pair comprising a source indicator from the plurality of indicators and the target metric, a second pair comprising an intermediate indicator from the plurality of indicators and the target metric, and a third pair comprising the source indicator and the intermediate indicator, and
compute a first predictivity value for the first pair, a second predictivity value for the second pair, and a third predictivity value for the third pair based on the machine learning model and the plurality of indicator pairs, wherein the first predictivity value indicates that the source indicator from the plurality of indicators is predictive of the target metric, wherein the second predictivity value indicates that the intermediate indicator from the plurality of indicators is predictive of the target metric, and wherein the third predictivity value indicates that the source indicator is predictive of the intermediate indicator; and
a graph component configured to generate a graph including the target metric and the plurality of indicators based on the first predictivity value, the second predictivity value, and the third predictivity value, wherein the graph includes a first edge between the source indicator and the target metric, a second edge between the intermediate indicator and the target metric, and a third edge between the source indicator and the intermediate indicator.
20 . The apparatus of claim 19 , further comprising:
a user interface configured to display a portion of the candidate time series data corresponding to the source indicator and the intermediate indicator based on the first predictivity value, the second predictivity value, and the third predictivity value.