Question 1
Choose 1
A luxury travel company wants to optimize its high-value customer retention strategy. The marketing team requires a predictive insight that estimates the total dollar amount a unified profile is likely to spend on bookings over the next 12 months. This prediction will be used to prioritize concierge service assignments.
Which native model type should a Data 360 Consultant select in Einstein Studio to predict this outcome?
Which native model type should a Data 360 Consultant select in Einstein Studio to predict this outcome?
Correct answer:
B. Regression model
Explanation:
The requirement is to predict a numeric dollar amount—the total future spend on bookings over the next 12 months. In Einstein Studio, this is a classic regression use case. Regression models are specifically designed to predict continuous numeric outcomes such as currency amounts, counts, or likelihood percentages .
Correct Option:
B. Regression model.
Regression models predict a number, such as a currency amount, count, or likelihood percentage. Salesforce documentation explicitly lists "Customer lifetime value of an account" and "Amount of an opportunity" as example use cases for regression models . The luxury travel company's goal of estimating total dollar spend over the next 12 months is a direct numeric prediction, making regression the correct native model type. The model is trained on historical booking data to learn patterns and produce a numeric prediction for each unified profile.
Incorrect Option:
A. Foundational model.
Foundational models (or foundation models) are large language models used for generative AI tasks like text generation, summarization, or chat completions. They are not designed for numeric regression predictions like future spend amounts .
C. Binary model.
Binary classification models predict outcomes with two possible values, such as true/false, yes/no, or won/lost. This model type cannot predict a dollar amount, which is a continuous numeric value rather than a two-option outcome .
D. Multiclass model.
Multiclass classification models predict outcomes with three to ten distinct categorical values, such as a tier or reason code. This is unsuitable for predicting a numeric dollar amount like future booking spend .
Reference:
Trailhead: Get to Know Einstein Studio; Salesforce Help: Evaluate Model Quality.
The requirement is to predict a numeric dollar amount—the total future spend on bookings over the next 12 months. In Einstein Studio, this is a classic regression use case. Regression models are specifically designed to predict continuous numeric outcomes such as currency amounts, counts, or likelihood percentages .
Correct Option:
B. Regression model.
Regression models predict a number, such as a currency amount, count, or likelihood percentage. Salesforce documentation explicitly lists "Customer lifetime value of an account" and "Amount of an opportunity" as example use cases for regression models . The luxury travel company's goal of estimating total dollar spend over the next 12 months is a direct numeric prediction, making regression the correct native model type. The model is trained on historical booking data to learn patterns and produce a numeric prediction for each unified profile.
Incorrect Option:
A. Foundational model.
Foundational models (or foundation models) are large language models used for generative AI tasks like text generation, summarization, or chat completions. They are not designed for numeric regression predictions like future spend amounts .
C. Binary model.
Binary classification models predict outcomes with two possible values, such as true/false, yes/no, or won/lost. This model type cannot predict a dollar amount, which is a continuous numeric value rather than a two-option outcome .
D. Multiclass model.
Multiclass classification models predict outcomes with three to ten distinct categorical values, such as a tier or reason code. This is unsuitable for predicting a numeric dollar amount like future booking spend .
Reference:
Trailhead: Get to Know Einstein Studio; Salesforce Help: Evaluate Model Quality.