Tableau-CRM-and-Einstein-Discovery-Consultant Exam Dumps - Salesforce Consultant Questions and Answers

In Einstein Analytics (Tableau CRM), the security options include:

• Inherited Security: Utilizes the existing Salesforce security model to control access.
• Row-level Security with Security Predicates: Offers granular control by restricting data visibility at the row level based on specific criteria.
• App Level Security: Manages access at the application level within Einstein Analytics.

Field-level security is not directly configurable within Einstein Analytics; instead, it's managed through the broader Salesforce security settings, affecting data visibility before it's imported into Einstein Analytics datasets. Reference: https://help.salesforce.com/s/articleView?id=sf.bi_security_overview.htm &type=5

Outliers in a dataset can significantly impact the results of a machine learning model, including those built by Einstein Discovery.expand_more Here's why changing binning methods is often the best approach:

• Outliers and Machine Learning: Outliers are extreme values that fall far outside the typical range of data.expand_more They can distort statistical summaries and skew the model's understanding of relationships within the data.
• Binning: Binning is the process of dividing continuous numerical data into discrete groups or bins.expand_more In Einstein Discovery, binning affects how data is visualized and analyzed.
• Fixed-width Binning: With fixed-width binning, you define equal-sized intervals for grouping data points. This can help mitigate the impact of outliers by spreading their influence across a wider range, reducing their effect on individual bins and the overall model.

Why other options are less ideal:

• A. Remove the outlier rows: While removing outliers can be effective, it risks discarding potentially valuable information. It's best used only when outliers are definitively incorrect.
• B. Nothing, because the field is not the outcome variable: Even predictor variables (inputs) with outliers can negatively impact the model. Outliers distort relationships the model is trying to learn.expand_more
• D. Remove the field because it has bad data: This is too drastic unless the whole field is unreliable. Binning adjustments are often enough to address outlier issues.