Skip to Content

Data Analyst Agent

Build an AI agent that analyzes data and creates visualizations.

What We’re Building

A data analyst agent that:

  • Loads and explores datasets
  • Performs statistical analysis
  • Creates charts and visualizations
  • Generates insights and reports

Quick Start

from isa_agent_sdk import Agent
 
agent = Agent(
    name="data-analyst",
    model="claude-sonnet-4-20250514",
    tools=["code_interpreter", "file_read", "database_query"],
    system_prompt="You are a data scientist. Analyze data and provide insights."
)
 
response = await agent.run("Analyze sales_data.csv and find the top performing products")

Implementation

from isa_agent_sdk import Agent
from isa_agent_sdk.tools import CodeInterpreter
 
# Configure with data science libraries
interpreter = CodeInterpreter(
    allowed_modules=[
        "pandas", "numpy", "matplotlib", "seaborn",
        "scipy", "sklearn"
    ]
)
 
agent = Agent(
    name="analyst",
    model="claude-sonnet-4-20250514",
    tools=[interpreter, "file_read", "database_query"],
    system_prompt="""You are an expert data analyst.
 
    When analyzing data:
    1. First explore the data structure
    2. Check for missing values and data quality
    3. Perform relevant statistical analysis
    4. Create visualizations to illustrate findings
    5. Summarize insights in plain language
    """
)
 
# Analyze a dataset
response = await agent.run("""
Load the file 'sales_2024.csv' and:
1. Show summary statistics
2. Find trends over time
3. Identify top customers
4. Create a visualization of monthly revenue
""")

Database Queries

from isa_agent_sdk import Agent
from isa_agent_sdk.tools import DatabaseQuery
 
# Configure database connection
db_tool = DatabaseQuery(
    connection_string="postgresql://user:pass@localhost/analytics",
    read_only=True  # Safety: prevent write operations
)
 
agent = Agent(
    name="sql-analyst",
    tools=[db_tool, "code_interpreter"],
    system_prompt="You can query the database and analyze results with Python."
)
 
response = await agent.run("What were our top 10 products by revenue last quarter?")

Visualization Output

The agent can create and save visualizations:

response = await agent.run("""
Create a dashboard with:
1. Revenue trend line chart
2. Product category pie chart
3. Customer segment bar chart
 
Save to 'dashboard.png'
""")
 
# Access generated files
for file in response.files:
    print(f"Generated: {file.path}")
Was this page helpful?
Home