New optimization metrics
Available in Beta
Selecting an optimization metric means that the best-performing model (according to that metric) will automatically be selected from among all models that are built during the training process.
The following optimization metrics have been added for binary models: Precision Rate and Recall Rate for the top quantity or percentile of entities. (Pre-existing options include Area Under the Curve, Accuracy and Logloss.)
Configure the Train-Test data split
Available in Beta
Now you can customize the cutoff point between the Train Set and Test Set in your training dataset.
Choosing the “split date” enables you to control the period and quantity of data used for training and tuning a model (Train Set), and for testing and evaluating a model (Test Set).
Salesforce connector
Available in Beta
Our new Salesforce connector allows you to import Salesforce data that can be used to train models and make predictions, and you can also export your predictions directly to Salesforce.
To connect to Salesforce, click Add connection under the “Connections” tab, select Salesforce, and follow the on-screen instructions.
Export predictions to Adjust and Singular
Available in Beta
Pecan now supports Adjust and Singular so marketing teams can optimize marketing campaigns by leveraging predictions right within their MMP platform.
To connect to Adjust or Singular, click Add connection under the “Connections” tab, select the relevant connector, and follow the on-screen instructions.
Easily edit connection settings
Available in Beta
You can now edit connection settings directly in the Pecan user interface. This is useful in cases where you need to update your credentials or move to a new server.
