Data Visualization with Langchain create_sql_agent

Question

So I was trying to write a code using Langchain to query my Postgres database and it worked perfectly then I tried to visualize the data if the user prompts like "Plot bar chart" now for that I found using python-repl tool could be an option but I am unable to use multiple tools (other toold is SQLDatabaseToolkit.

toolkit = SQLDatabaseToolkit(db=db, llm=llm)

agent_executor = create_sql_agent(
    llm=llm,
    toolkit=toolkit,
    agent_type=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
    input_variables = ["question", "agent_scratchpad"],
    suffix = suffix, # must have history as variable,
    verbose = True,
    handle_parsing_errors=True
)

This is my code that can query things from my database, so how can I plot from this queries?

Answer 1

You can create two or more agents and use them as tools with initialize_agent().

Example:- Consider CSV file having data:-

question, answer
2+3,
22/7,
9+2,

I want to fetch data from CSV file (database in your case), then use PythonREPLTool to process data and save it back to the CSV file. For this, I can create two agents as follows:-

Step 0:- Import dependencies

from langchain.llms import OpenAI

from langchain.agents.agent_toolkits import create_python_agent
from langchain.agents.agent_toolkits import create_csv_agent
from langchain.agents import initialize_agent
from langchain.tools.python.tool import PythonREPLTool

Step 1:- Create Python agent

llm = OpenAI(model_name="text-davinci-003", temperature=0)

python_agent = create_python_agent(llm, tool=PythonREPLTool(), verbose=True)

Step 2:- Create CSV agent

current_dir = os.path.dirname(__file__)
data_file_path = os.path.join(current_dir, "data.csv")
csv_agent = create_csv_agent(llm, path=data_file_path, verbose=True)

Step 3:- Create a new agent to use Python and CSV agents as tool

agent = initialize_agent(
    tools=[
        Tool(
            name="PythonAgent",
            func=python_agent.run,
            description="""Useful to run python commands""",
        ),
        Tool(
            name="CSVAgent",
            func=csv_agent.run,
            description="""Useful to manipulate csv files""",
        ),
    ],
    llm=llm,
    agent_type=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
    verbose=True,
)
output = agent.run(
    "In data.csv file, evaluate expression in first column and save it to second cloumn. Do this for all rows and save new file as solved.csv"
)

Output:-

,question,answer
0,2+3,5.0
1,22/7,3.142857142857143
2,9+2,11.0