Prediction types and dataset requirements for models
Choosing the numeric, binary classification or multiclass classification use case from the outcome column, and shaping a dataset that meets Einstein Discovery's row, column, known-outcome and granularity requirements before you create a model.
Build a model by assessing data and selecting one of the three types of predictions (numeric, binary, multi-classification).
~8 min read · 8 practice questions ·
What the exam expects
- Pick the numeric, binary classification or multiclass classification use case from the outcome column's data type and number of values.
- Set the model goal correctly: Maximize or Minimize for numeric and binary outcomes, the most likely class for multiclass.
- Check a dataset against the minimums (rows with a known outcome, columns) and maximums before creating a model.
- Explain which rows Einstein Discovery trains on and why rows without an outcome drop out.
- Shape data from one or more sources into rows at the granularity the business question needs.
Key terms: outcome variable (the column the model predicts, set in I Want to Analyze); explanatory variable or predictor (a column that may influence the outcome); observation (one row of the dataset); use case (numeric, binary classification or multiclass classification, set by the outcome); granularity (what one row represents, such as a customer, a day or a shipment).
Key facts
- Detail
- Numeric, binary classification, multiclass classification
- Detail
- Text outcome with 3-10 possible classes
- Detail
- Numeric and binary outcomes only
- Detail
- Text values; convert 1 and 0 to text
- Detail
- 400 minimum with a known outcome, 20,000,000 maximum
- Detail
- 3 minimum (outcome plus 2), 50 maximum
- Detail
- Omitted from analysis
- Detail
- Field needs values in at least 400 rows
- Detail
- Creator needs access to at least 400 rows
- Detail
- Live datasets; multivalue fields
Sources for this section (8)
- Define Your Target Outcomehelp.salesforce.com (opens in a new tab)
- About Modelshelp.salesforce.com (opens in a new tab)
- Configure Modelhelp.salesforce.com (opens in a new tab)
- Address Common Data Issueshelp.salesforce.com (opens in a new tab)
- Capacities and Requirementshelp.salesforce.com (opens in a new tab)
- Determine Data Requirementshelp.salesforce.com (opens in a new tab)
- Select a Datasethelp.salesforce.com (opens in a new tab)
- Create a Model from a Datasethelp.salesforce.com (opens in a new tab)
- Why might a text column you expect be missing from the I Want to Analyze list?
- When does a model pick up new rows added to its dataset after creation?
- Which algorithm family is offered for numeric and binary models but not multiclass?
- What does Show as whole number >= 0 do?
- When should a proof of concept limit its input variables?
The questions are free to read. Answers with their sources and self-grading are included with the pack.
Keep reading with the CRM Analytics and Einstein Discovery Consultant pack
- Core concepts
- Exam traps
- Worked scenario
The pack includes:
- All 26 lessons of the CRM Analytics and Einstein Discovery Consultant guide, with reading progress
- Interactive decision walks and step-by-step sequence steppers
- Practice exams purchase adds inline bank questions and topic practice (8 questions for this lesson)
- An official source cited for every claim
- An AI tutor that answers from the guide
- Choose guide-only, practice-only, or save $4 with both