Predictive Modeling Validation Metrics Dataset (version 2022.4)

The ChannelMix Predictive Modeling Validation Metrics Dataset is used to power the Validation Metrics section of the Pipeline with Predictive Modeling Dashboard. The dataset contains metrics produced while validating the quality of the trained model..

The dataset is refreshed at the same cadence as the Predictive Model is refreshed - monthly by default. Only the most recently produced results are displayed in the dashboard.

Field Name

Field Type

Description

Report Date

DATE

The date being used to compare predicted vs. actual.

Predicted Values

DOUBLE PRECISION

The predicted conversions from the model for this report date.

Actual Values DOUBLE PRECISION The actual number of conversions that occurred on this report date. 
Lower Prediction DOUBLE PRECISION Lower bound of the 90% HPDI region for predicting the target. A conservative estimate of the smallest value the target will take in real life.
Upper Predictions DOUBLE PRECISION Upper bound of the 90% HPDI region for predicting the target. A conservative estimate of the largest value the target will take in real life.
Spend DOUBLE PRECISION The amount spent on this feature.
Feature TEXT Name of the feature in the model. I.e. Paid Social, Display, etc.
Pred Actual DOUBLE PRECISION The difference between the predicted values and actual values.
R Squared DOUBLE PRECISION The R-squared metric for the model.
Mean Squared Error DOUBLE PRECISION The Mean-Squared Error for the model.
SMAPE DOUBLE PRECISION The sMAPE metric for the model.
Intercept DOUBLE PRECISION 1 - R2
Model ID TEXT ID to identify the trained model used for optimization.
Target Variable Pretty TEXT
A more descriptive name for the target variable to be displayed in the dashboard.
Target Variable TEXT The name of the field the model is trained to predict.
OneView Name TEXT The name of the OneView data is pulled from to model.
Feature Type TEXT The name of the feature used to predict the target variable.
Conversion Name TEXT The name of the field the model is trained to predict.
Model Run Date DATE Date the model was trained.
Training Start Date TIMESTAMP Earliest date provided to the model in training.
Training End Date TIMESTAMP Most recent date provided to the model in training.
Min Daily Cost DOUBLE PRECISION Smallest amount spent in a single day summed across all channels.
Max Daily Cost DOUBLE PRECISION Largest amount spent in a single day summed across all channels.
Insert Date TIMESTAMP The date in which the record was loaded into the data warehouse
Channelmix Profile TEXT The ChannelMix Profile associated to the data
Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request

Comments

0 comments

Please sign in to leave a comment.