Praison AI

Pandas Agent

Pandas data manipulation tools for AI agents.

python
from praisonaiagents import Agent
from praisonaiagents import read_csv, get_summary

agent = Agent(
    name="Data Analyst",
    instructions="Analyse tabular data with pandas tools.",
    tools=[read_csv, get_summary],
)
agent.start("Describe the sales.csv dataset and highlight outliers")

The user names a dataset; the agent loads it with pandas tools and explains key statistics.

Diagram

Input

Agent

Tool

Output

Diagram
Pandas AgentAgentUserPandas AgentAgentUserRequestProcess requestResultResponse

Pandas Tools

Use Pandas Tools to manipulate and analyze data with AI agents.

  1. Install Dependencies

    First, install the required packages:

    bash
    pip install praisonaiagents praisonai-tools pandas
  2. Import Components

    Import the necessary components:

    python
    from praisonaiagents import Agent, Task, AgentTeam
from praisonai_tools import read_csv, read_excel, write_csv, write_excel, filter_data, get_summary, group_by, pivot_table
  3. Create Agent

    Create a data analysis agent:

    python
    data_agent = Agent(
    name="DataAnalyst",
    role="Data Analysis Specialist",
    goal="Analyze and transform data efficiently.",
    backstory="Expert in data manipulation and statistical analysis.",
    tools=[read_csv, read_excel, write_csv, write_excel, filter_data, get_summary, group_by, pivot_table],
    reflection=False
)
  4. Define Task

    Define the analysis task:

    python
    analysis_task = Task(
    description="Analyze sales data to identify trends and patterns.",
    expected_output="Statistical summary and key insights.",
    agent=data_agent,
    name="sales_analysis"
)
  5. Run Agent

    Initialize and run the agent:

    python
    agents = AgentTeam(
    agents=[data_agent],
    tasks=[analysis_task],
    process="sequential"
)
agents.start()

Understanding Pandas Tools

What are Pandas Tools?

Pandas Tools provide data analysis capabilities for AI agents: – Data filtering and selection – Statistical analysis – Data transformation – Group operations – Pivot table creation

Key Components

Data Agent

Create specialized data agents:

python
Agent(tools=[read_csv, read_excel, write_csv, write_excel, filter_data, get_summary, group_by, pivot_table])
 Analysis Task

Define analysis tasks:

python
Task(description="analysis_query")
 Process Types

Sequential or parallel processing:

python
process="sequential"
 Analysis Options

Customize analysis parameters:

python
group_by="category", agg_func="mean"

Available Functions

python
from praisonai_tools import read_csv
from praisonai_tools import read_excel
from praisonai_tools import write_csv
from praisonai_tools import write_excel
from praisonai_tools import filter_data
from praisonai_tools import get_summary
from praisonai_tools import group_by
from praisonai_tools import pivot_table

Examples

Basic Data Analysis Agent

python
from praisonaiagents import Agent, Task, AgentTeam
from praisonai_tools import read_csv, filter_data, get_summary, group_by

# Create data analysis agent
data_agent = Agent(
    name="DataExpert",
    role="Data Analyst",
    goal="Process and analyze data efficiently.",
    backstory="Expert in data analysis and statistical methods.",
    tools=[read_csv, filter_data, get_summary, group_by],
    reflection=False
)

# Define analysis task
analysis_task = Task(
    description="Analyze sales performance data.",
    expected_output="Sales analysis report.",
    agent=data_agent,
    name="sales_analysis"
)

# Run agent
agents = AgentTeam(
    agents=[data_agent],
    tasks=[analysis_task],
    process="sequential"
)
agents.start()

Advanced Data Operations with Multiple Agents

python
from praisonaiagents import Agent, Task, AgentTeam
from praisonai_tools import read_csv, read_excel, write_csv, write_excel, filter_data, get_summary, group_by, pivot_table

# Create data import/export agent
io_agent = Agent(
    name="DataIO",
    role="Data IO Specialist",
    goal="Handle data import and export operations.",
    tools=[read_csv, read_excel, write_csv, write_excel],
    reflection=False
)

# Create analysis agent
analysis_agent = Agent(
    name="Analyzer",
    role="Data Analysis Specialist",
    goal="Perform complex data analysis.",
    tools=[filter_data, get_summary, group_by, pivot_table],
    reflection=False
)

# Define tasks
io_task = Task(
    description="Import and export data",
    agent=io_agent,
    name="data_io"
)

analysis_task = Task(
    description="Analyze data",
    agent=analysis_agent,
    name="data_analysis"
)   

# Run agents
agents = AgentTeam(
    agents=[io_agent, analysis_agent],
    tasks=[io_task, analysis_task],
    process="sequential"
)
agents.start()

Best Practices

Agent Configuration

Configure agents with clear analysis focus:

python
from praisonai_tools import read_csv, read_excel, write_csv, write_excel, filter_data, get_summary, group_by, pivot_table

Agent(
    name="DataScientist",
    role="Data Analysis Specialist",
    goal="Extract meaningful insights from data",
    tools=[read_csv, read_excel, write_csv, write_excel, filter_data, get_summary, group_by, pivot_table]
)
Task Definition

Define specific analysis objectives:

python
Task(
    description="Analyze sales trends and forecast future growth",
    expected_output="Growth analysis with forecasts"
)

Common Patterns

Data Analysis Pipeline

python
from praisonaiagents import Agent, Task, AgentTeam
from praisonai_tools import read_csv, read_excel, write_csv, write_excel, filter_data, get_summary, group_by, pivot_table

# Data preparation agent
prep_agent = Agent(
    name="Preparer",
    role="Data Preparation",
    tools=[read_csv, read_excel, write_csv, write_excel, filter_data, get_summary, group_by, pivot_table]
)

# Analysis agent
analyst = Agent(
    name="Analyst",
    role="Data Analyst"
)

# Define tasks
prep_task = Task(
    description="Clean and prepare data",
    agent=prep_agent
)

analysis_task = Task(
    description="Analyze prepared data",
    agent=analyst
)

# Run workflow
agents = AgentTeam(
    agents=[prep_agent, analyst],
    tasks=[prep_task, analysis_task]
)

Related

Custom Tools

Build your own agent tools

 Tools Overview

Browse PraisonAI tool documentation