A consultant is preparing a dataset to predict customer lifetime value and is collecting data from a questionnaire that asks for demographic information. A very small number of respondents fill in the Income box, but the consultant thinks that it is an informative column even though it only represents 1% of respondents.
What should the consultant do?
consultant is reviewing a model that is set to maximize the daily sales quantity of consumer products in stores, and they see this recommendation.
Which action should the consultant take?
Upon reviewing the data model and noticing the high correlation alert between 'Store' and daily sales quantity, the appropriate action is to verify with the client their expectations regarding the influence of the Store field on daily sales. Here's the rationale:
Understanding the Role of 'Store' in the Model: Before making any changes to the model, it's crucial to understand whether the 'Store' field is expected to be a strong predictor based on the business context. If the client expects that different stores inherently have different sales volumes due to factors like location, size, or customer base, this correlation may be both meaningful and desired.
Potential Data Leakage: High correlation warnings can sometimes indicate data leakage, where a predictor (like 'Store') might inadvertently include information about the outcome variable (daily sales quantity). It's essential to verify whether this correlation makes sense logically or if it's skewing the model predictions.
Client Consultation: Consulting with the client helps ensure that any modeling decisions align with their business knowledge and expectations. It's about validating the model against real-world expectations and ensuring it remains a useful tool for decision-making.
By taking these steps, the consultant not only adheres to best practices in data science by validating model inputs and their implications but also ensures that the model aligns with the client's business strategies and operational realities.
Universal Containers wants to create two dashboards and has two user groups. The 'Regional Performance' dashboard should be accessible to sales reps and managers/executives to keep track of
how sales reps are performing in each region. Sales reps must only be able to see data pertaining to their respective region. The 'National Performance' dashboard is using the same data as the other
dashboard but should only be accessible to managers/executives to compare data across all regions.
In addition to row-level security to view only regional data, how should a consultant ensure that sales reps are unable to view the 'National Performance' dashboard?
The marketing team at Cloud Kicks has five dashboards in an app. Four widgets are replicas of each other in three of the dashboards.
What is the best way to maintain these widgets?
To maintain consistency and ease of updates across multiple dashboards, creating or editing a component for the widgets is the most effective method. This approach:
Efficiency in Updates: Allows changes to be made in one place, which automatically propagates to all instances where the component is used across dashboards.
Consistency: Ensures uniformity in the appearance and functionality of the widgets across different dashboards.
Simplicity: Reduces the need for redundant work, where each widget would otherwise need to be updated individually.
A consultant runs the sharing inheritance coverage assessment for the Opportunity object and finds that some records exceed 400 sharing descriptors.
What should the consultant do?
When a record exceeds 400 sharing descriptors, it can cause performance issues or sharing rule complications in CRM Analytics. In such cases, the recommended solution is to use security predicates, which allow fine-tuned control over which data is visible to users based on their sharing rules and permissions. Security predicates reduce the number of sharing descriptors by enforcing security at the dataset level rather than relying solely on record-sharing mechanisms.
Increasing the sharing descriptor limit is not an available option, and Salesforce Support does not typically increase this limit, making the use of security predicates the best approach.
