Customizing model in Einstein Discovery

In Create a model in Einstein Discovery Part 1 – Mortgage Banking Technology | Origination Strategy Consulting (takefiveconsulting.com) and CREATE A MODEL IN EINSTEIN DISCOVERY PART 2 – Mortgage Banking Technology | Origination Strategy Consulting (takefiveconsulting.com), we have covered how to create a model from a dataset in Einstein Discovery. In this article, we will discuss about customizing the created model in Einstein Discovery.

To get started customizing your model, click Settings.

1. Dataset:
   • A dataset consists of rows and columns, where each row represents an observation and each column signifies a variable.
2. Variables Table:
   • In the Variables Table, you find the variables present in the model:
     • The first variable, “Converted,” serves as the outcome variable, representing the business outcome to enhance, with the goal of maximizing it.
     • Following are the explanatory variables, which are explored to understand their influence on the outcome variable.
     • Importance measures the relative impact of a variable on the model’s predicted outcome, quantified as a percentage. Higher percentages indicate greater influence, considering variable interactions. It also accounts for correlations and selects the most informative variables for prediction.
     • The column dropdown enables switching to view correlations, indicating the statistical association between explanatory variables and the outcome. Correlation strength is represented as a percentage, with higher values indicating stronger relationships. This measure helps assess how well each field independently predicts the outcome.
     • Data alerts prompt attention if Einstein Discovery identifies potential data issues requiring special consideration.
3. General Settings:
   • General Settings provide insights into the dataset utilized by the model. Additionally, you can view and modify validation and algorithm settings for the model.

Edit Variable Settings

In the variables table, click Lead Source. In the right panel, configure settings for the selected variable.

• Enable the “Analyze for bias” option if there are concerns about potential bias in the variable, activating Einstein Discovery’s bias detection features.

• Choose from available Transform options to manipulate variable values during analysis. Transformation options include fuzzy matching, sentiment detection, text clustering, and replacing missing values. These transformations affect only the model’s data, leaving the original dataset values unchanged. For instance, fuzzy matching corrects minor typographical discrepancies in text values, enhancing the model’s categorization accuracy and predictive capabilities.
• The “Include Only” feature displays variable-associated values, starting with the most frequent one. By deselecting a value, you can instruct Einstein to either exclude it from analysis or merge it into the “Other” category.
• The Histogram visualizes the frequency distribution of values within the dataset.

To learn more about models, click About Models (salesforce.com)

We hope this article helped you in customizing model in Einstein Discovery.