Prediction events

Note: Your integration may add custom parameters. The tables only list parameters that exist in Synerise and SDKs/APIs by default.

snr.prediction.score

A prediction was completed for a profile. Check the parameters to see the prediction score.

Default retention: 30 days

Parameter	Type	Example	Description
Base parameters	various	n/a	Parameters that exist in all events
modelId	string	`bftjpcmdxrku`	Unique ID of the prediction. It is part of the prediction's URL.
modelName	string	`ltv_prediction`	Name entered when creating the prediction. You can use it to summarize what the prediction does or give the result an identifier that can be used later for another purpose.
score_label	string	`Very low`	A textual label of the score, on a 2- or 5-point scale depending on the prediction's configuration.
score	number	`101.297`	The score as a number
percentile	number	`39`	The percentile where this score is located among all other scores of the prediction.
configurationVersion	number	`3`	Allows to verify whether two consecutive calculations ran with the same settings (only for debugging).
modelVersion	string	`0.285`	Model version allows to verify whether two consecutive calculations ran on the same model parameters (only for debugging).
featureInfluence	array		Local feature importance
classes	array of strings	`["0","1"]`	Available classes. Only exists in classification predictions.
prediction	string	`1`	Predicted class. Only exists in classification predictions.

A propensity prediction was completed for a profile. Check the parameters to see the prediction score.

Default retention: 30 days

Parameter	Type	Example	Description
Base parameters	various	n/a	Parameters that exist in all events
modelId	string	`bftjpcmdxrku`	Unique ID of the prediction. It is part of the prediction's URL.
modelName	string	`ltv_prediction`	Name entered when creating the prediction. You can use it to summarize what the prediction does or give the result an identifier that can be used later for another purpose.
score_label	string	`Very low`	A textual label of the score, on a 2- or 5-point scale depending on the prediction's configuration.
scoreName	string	`ltv_prediction`	Deprecated parameter: custom score name entered when creating the prediction. You can use it to summarize what the prediction does or give the result an identifier that can be used later for another purpose.
score	number	`101.297`	The score as a number
percentile	number	`39`	The percentile where this score is located among all other scores of the prediction.
configurationVersion	number	`3`	Allows to verify whether two consecutive calculations ran with the same settings (only for debugging).
modelVersion	string	`0.285`	Model version allows to verify whether two consecutive calculations ran on the same model parameters (only for debugging).

A lookalike prediction was completed for a profile. Check the parameters to see the prediction score.

Default retention: 30 days

Parameter	Type	Example	Description
Base parameters	various	n/a	Parameters that exist in all events
modelId	string	`bftjpcmdxrku`	Unique ID of the prediction. It is part of the prediction's URL.
modelName	string	`ltv_prediction`	Name entered when creating the prediction. You can use it to summarize what the prediction does or give the result an identifier that can be used later for another purpose.
score_label	string	`Very low`	A textual label of the score, on a 2- or 5-point scale depending on the prediction's configuration.
scoreName	string	`ltv_prediction`	Deprecated parameter: custom score name entered when creating the prediction. You can use it to summarize what the prediction does or give the result an identifier that can be used later for another purpose.
score	number	`101.297`	The score as a number
percentile	number	`39`	The percentile where this score is located among all other scores of the prediction.
configurationVersion	number	`3`	Allows to verify whether two consecutive calculations ran with the same settings (only for debugging).
modelVersion	string	`0.285`	Model version allows to verify whether two consecutive calculations ran on the same model parameters (only for debugging).

A time optimizer prediction was made for a profile. Check the parameters to see the result.

Default retention: 30 days

Parameter	Type	Example	Description
Base parameters	various	n/a	Parameters that exist in all events
isFallback	boolean	`false`	This parameter can take two values: `true` or `false`. It describes whether there was enough data to make a prediction. If there was not enough data, this parameter takes `true` value. In such case, the system takes the most used time from the distribution of other profiles.
nextNhours	integer	`12`	This parameter is the value used in the Custom time period option.
expireAt	string	`2022-12-04T19:39:47.398318+00:00`	This parameter contains the result of the prediction: the best date and time proposed by the algorithm given in UTC.
topN	integer	`1`	This parameter takes numerical values and it informs how many best hour predictions matched the conditions and exclusions. Usually the value is `1`.
mode	string	`standard`	This parameter returns the name of the mode used for calculating the prediction.
timeIndexExclude	array of arrays of integers	`[[166,167]]`	This parameter returns the days and hours that are excluded from the time proposed by the algorithm.
predictedHour	number	`163`	This parameter returns the predicted hour in the form of an hour of the week (for example, according to the example in the `expireAt` description, the value for the `predictedHour` parameter is 163 - Sunday, 7 P.M. UTC)