Correlation Finder Agent

Agent Description:

The Correlation Finder Agent automatically identifies and explains relationships between variables within datasets. It fetches data through APIs, cleans it, and calculates correlation coefficients such as Pearson and Spearman. The agent then highlights the strongest positive and negative correlations, explaining them in clear, plain language. Finally, it provides insights and optional visualizations (such as heatmaps) to support data-driven decision-making.

Purpose and Components
  • Purpose: To automatically detect and interpret relationships between variables in structured datasets. The agent computes correlation coefficients, visualizes the correlation matrix, and extracts key patterns to support data-driven decision-making.
  • Components:

    • GET Data Connector for fetching datasets via API requests.

    • Data Processing Engine for cleaning, validating, and preparing datasets for analysis.

    • Analytical Core for computing Pearson/Spearman correlations across numerical variables.

    • Visualization Module for generating graphical correlation matrices and summaries.

    • Insight Generator for identifying and explaining the strongest relationships between key variables.

Supported Capabilities

  • Pull structured data from APIs using GET requests.

  • Validate and preprocess datasets (remove nulls, standardize data types).

  • Compute pairwise correlations between numerical variables.

  • Identify top positive and negative correlations with significance levels.

  • Generate visual correlation matrices (heatmaps, tables).

  • Summarize and interpret insights in plain language.

  • Return both matrix and insights directly in chat for user understanding.

LLM Used

  • GPT-5 (Default, inherits capabilities from parent orchestration agent)

    Note: To learn more about the LLM and to modify its behavior, refer to the Configuring LLM settings section.

Sub-Agents

1. Data Extracting Agent

  • Role:Data Fetcher
  • Scope:Retrieve and prepare dataset from GET API for analysis.
  • Description: This sub-agent connects to an API endpoint via GET requests to fetch structured business data (JSON/CSV). It performs schema validation, data type normalization, and null handling to ensure data consistency. The cleaned dataset is then passed to the analytical sub-agent for further processing.
  • LLM Used: Default (inherits from parent).
  • Tools used: GET Tool

2. Correlation Analyzer Agent

  • Role:Pattern Finder
  • Scope:Compute correlations and highlight significant variable relationships.
  • Description: This sub-agent receives the prepared dataset from DataFetcherAgent and performs pairwise correlation analysis using statistical methods (Pearson/Spearman). It generates a correlation matrix and identifies top positive and negative correlations. The agent then explains the key insights in plain English and returns them as chat output along with a tabular or visual summary.
  • LLM Used: Default (inherits from parent).
Tools Used:

  • GET Tool: Fetches external data from APIs for analysis.

  • Data Cleaning Utility: Removes inconsistencies, missing values, and invalid entries.

Note: For details on modifying the Tools, refer Tools Library section.
Agent Workflow Behavior Summary

  1. DataFetcherAgent retrieves dataset from API endpoint using a GET request.

  2. It cleans, validates, and formats the data for correlation analysis.

  3. CorrelationAnalyzerAgent computes the correlation matrix between all numerical variables.

  4. It extracts and summarizes key insights on variable relationships.

  5. The main agent compiles results, presents them in chat, and displays the correlation matrix.
Sample Questions:

  • Find correlations between marketing spend, sales revenue, and customer satisfaction.

  • Show me which variables are most related to customer retention.

  • Generate a correlation matrix for my performance dataset.

  • Summarize the strongest positive and negative correlations in this data.

  • Which factors most influence revenue growth in the dataset