Chapter 14 : Custom Einstein Predictions

 There is Einstein for Sales/Service/Marketing clouds.

Einstein gives you the ability to do your own custom predictions.

Sign up for prediction builder org using https://sfdc.co/PB-org 

Example usecase:

Predict who is going to pay their bills late so that we can reachout to them and ensure payments come on time.

Steps:

Click on Get Started


click on new prediction


give it a name


You can make predictions on standard or custom objects

You can focus on segment of data or all records of the selected object. For example we can ignore autopay customers who will never pay late.


Select a field to predict.

Uses historical data and learns from that data and finds the patterns.

In this example, Einstein would learn from all invoices which are already paid and will make predictions on invoices which are in pending state. Click on Check Data to verify that we have enough amount of data in org for Einstein to learn. Einstein requires atleast 400 records to learn from. It needs atleast 100 yes and atleast 100 No.


Select fields from which Einstein should learn from. By default it would select all the fields of the object.

Einstein would create a new custom field to store prediction results. It would be a percentage probabilistic value(0-100).
click on "Build My Prediction". It would create a record in predictions. Review the score card.



click on predictions tab to dig deeper into what Einstein learnt


Once reviewed, click on Enabled in record. Einstein will now begin scoring.

Those with highest value of prediction score are likely to pay bills late in this example so that we can focus on those.

Setup VSCode with Salesforce

Step by Step guide to setup VSCode with Salesforce Install vscode Install Salesforce CLI (developer.salesforce.com/tools/sfdxcli) Perform fo...