Task Solving with Code Generation, Execution and Debugging¶
In this notebook, we demonstrate how to use AssistantAgent and UserProxyAgent to write code and execute the code. Here AssistantAgent is an LLM-based agent that can write Python code (in a Python coding block) for a user to execute for a given task. UserProxyAgent is an agent which serves as a proxy for the human user to execute the code written by AssistantAgent, or automatically execute the code. Depending on the setting of human_input_mode and max_consecutive_auto_reply, the UserProxyAgent either solicits feedback from the human user or returns auto-feedback based on the result of code execution (success or failure and corresponding outputs) to AssistantAgent. AssistantAgent will debug the code and suggest new code if the result contains error. The two agents keep communicating to each other until the task is done.
{=mdx}
:::info Requirements
Install the following packages before running the code below:
```bash
pip install pyautogen matplotlib yfinance
```
For more information, please refer to the [installation guide](/docs/installation/).
:::
In [2]:
from IPython.display import Image, display
import autogen
from autogen.coding import LocalCommandLineCodeExecutor
config_list = autogen.config_list_from_json(
"OAI_CONFIG_LIST",
filter_dict={"tags": ["gpt-4"]}, # comment out to get all
)
# When using a single openai endpoint, you can use the following:
# config_list = [{"model": "gpt-4", "api_key": os.getenv("OPENAI_API_KEY")}]
Example Task: Check Stock Price Change¶
In the example below, let's see how to use the agents in AutoGen to write a python script and execute the script. This process involves constructing a AssistantAgent to serve as the assistant, along with a UserProxyAgent that acts as a proxy for the human user. In this example demonstrated below, when constructing the UserProxyAgent, we select the human_input_mode to "NEVER". This means that the UserProxyAgent will not solicit feedback from the human user. It stops replying when the limit defined by max_consecutive_auto_reply is reached, or when is_termination_msg() returns true for the received message.
In [3]:
# create an AssistantAgent named "assistant"
assistant = autogen.AssistantAgent(
name="assistant",
llm_config={
"cache_seed": 41, # seed for caching and reproducibility
"config_list": config_list, # a list of OpenAI API configurations
"temperature": 0, # temperature for sampling
}, # configuration for autogen's enhanced inference API which is compatible with OpenAI API
)
# create a UserProxyAgent instance named "user_proxy"
user_proxy = autogen.UserProxyAgent(
name="user_proxy",
human_input_mode="NEVER",
max_consecutive_auto_reply=10,
is_termination_msg=lambda x: x.get("content", "").rstrip().endswith("TERMINATE"),
code_execution_config={
# the executor to run the generated code
"executor": LocalCommandLineCodeExecutor(work_dir="coding"),
},
)
# the assistant receives a message from the user_proxy, which contains the task description
chat_res = user_proxy.initiate_chat(
assistant,
message="""What date is today? Compare the year-to-date gain for META and TESLA.""",
summary_method="reflection_with_llm",
)
user_proxy (to assistant): What date is today? Compare the year-to-date gain for META and TESLA. -------------------------------------------------------------------------------- assistant (to user_proxy): First, let's get the current date using Python. ```python # Python code from datetime import date # Get today's date today = date.today() # Print today's date print("Today's date:", today) ``` Next, we need to fetch the stock prices for META (Facebook) and TESLA for the current year. We can use the `yfinance` library in Python to fetch this data. If `yfinance` is not installed, it can be installed using pip: `pip install yfinance`. Here is the Python code to fetch the stock prices and calculate the year-to-date gain: ```python # Python code import yfinance as yf from datetime import datetime # Get the current year current_year = datetime.now().year # Download stock data for the current year meta_data = yf.download('FB', start=f'{current_year}-01-01', end=today) tesla_data = yf.download('TSLA', start=f'{current_year}-01-01', end=today) # Calculate the year-to-date gain for each stock meta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100 tesla_ytd_gain = ((tesla_data['Close'][-1] - tesla_data['Close'][0]) / tesla_data['Close'][0]) * 100 # Print the year-to-date gain for each stock print(f'META year-to-date gain: {meta_ytd_gain}%') print(f'TESLA year-to-date gain: {tesla_ytd_gain}%') ``` This code fetches the closing prices for META and TESLA for the current year, calculates the year-to-date gain for each stock, and prints the results. The year-to-date gain is calculated as the percentage change in the closing price from the first trading day of the year to the most recent trading day. -------------------------------------------------------------------------------- >>>>>>>> EXECUTING 2 CODE BLOCKS (inferred languages are [python, python])... user_proxy (to assistant): exitcode: 1 (execution failed) Code output: Today's date: 2024-04-12 Traceback (most recent call last): File "/Users/ekzhu/autogen/notebook/coding/tmp_code_cb9ef30baa23cf28e127198c0ebeb7e6.py", line 9, in <module> meta_data = yf.download('FB', start=f'{current_year}-01-01', end=today) ^^^^^ NameError: name 'today' is not defined -------------------------------------------------------------------------------- assistant (to user_proxy): I apologize for the oversight. The variable `today` was defined in the first code block but not in the second one. Let's correct this by defining `today` in the second code block as well. Here's the corrected code: ```python # Python code import yfinance as yf from datetime import datetime, date # Get the current year and today's date current_year = datetime.now().year today = date.today() # Download stock data for the current year meta_data = yf.download('FB', start=f'{current_year}-01-01', end=today) tesla_data = yf.download('TSLA', start=f'{current_year}-01-01', end=today) # Calculate the year-to-date gain for each stock meta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100 tesla_ytd_gain = ((tesla_data['Close'][-1] - tesla_data['Close'][0]) / tesla_data['Close'][0]) * 100 # Print the year-to-date gain for each stock print(f'META year-to-date gain: {meta_ytd_gain}%') print(f'TESLA year-to-date gain: {tesla_ytd_gain}%') ``` This code should now correctly fetch the stock prices for META and TESLA, calculate the year-to-date gain for each, and print the results. -------------------------------------------------------------------------------- >>>>>>>> EXECUTING CODE BLOCK (inferred language is python)... user_proxy (to assistant): exitcode: 1 (execution failed) Code output: [*********************100%%**********************] 1 of 1 completed 1 Failed download: ['FB']: Exception('%ticker%: No timezone found, symbol may be delisted') /Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/yfinance/utils.py:775: FutureWarning: The 'unit' keyword in TimedeltaIndex construction is deprecated and will be removed in a future version. Use pd.to_timedelta instead. df.index += _pd.TimedeltaIndex(dst_error_hours, 'h') [*********************100%%**********************] 1 of 1 completed /Users/ekzhu/autogen/notebook/coding/tmp_code_9ca584c3d8b484c70ee49c1ce5e4b975.py:14: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]` meta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100 Traceback (most recent call last): File "/Users/ekzhu/autogen/notebook/coding/tmp_code_9ca584c3d8b484c70ee49c1ce5e4b975.py", line 14, in <module> meta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100 ~~~~~~~~~~~~~~~~~~^^^^ File "/Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/pandas/core/series.py", line 1109, in __getitem__ return self._values[key] ~~~~~~~~~~~~^^^^^ IndexError: index -1 is out of bounds for axis 0 with size 0 -------------------------------------------------------------------------------- assistant (to user_proxy): I apologize for the confusion. It seems like the ticker symbol for Meta Platforms Inc. (formerly Facebook) has changed from 'FB' to 'META'. Let's correct this in the code. Also, it seems like the data for the current day might not be available yet, so let's fetch the data until the previous day. Here's the corrected code: ```python # Python code import yfinance as yf from datetime import datetime, timedelta # Get the current year and yesterday's date current_year = datetime.now().year yesterday = datetime.now() - timedelta(1) # Download stock data for the current year meta_data = yf.download('META', start=f'{current_year}-01-01', end=yesterday) tesla_data = yf.download('TSLA', start=f'{current_year}-01-01', end=yesterday) # Calculate the year-to-date gain for each stock meta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100 tesla_ytd_gain = ((tesla_data['Close'][-1] - tesla_data['Close'][0]) / tesla_data['Close'][0]) * 100 # Print the year-to-date gain for each stock print(f'META year-to-date gain: {meta_ytd_gain}%') print(f'TESLA year-to-date gain: {tesla_ytd_gain}%') ``` This code should now correctly fetch the stock prices for META and TESLA, calculate the year-to-date gain for each, and print the results. -------------------------------------------------------------------------------- >>>>>>>> EXECUTING CODE BLOCK (inferred language is python)... user_proxy (to assistant): exitcode: 0 (execution succeeded) Code output: /Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/yfinance/utils.py:775: FutureWarning: The 'unit' keyword in TimedeltaIndex construction is deprecated and will be removed in a future version. Use pd.to_timedelta instead. df.index += _pd.TimedeltaIndex(dst_error_hours, 'h') [*********************100%%**********************] 1 of 1 completed /Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/yfinance/utils.py:775: FutureWarning: The 'unit' keyword in TimedeltaIndex construction is deprecated and will be removed in a future version. Use pd.to_timedelta instead. df.index += _pd.TimedeltaIndex(dst_error_hours, 'h') [*********************100%%**********************] 1 of 1 completed /Users/ekzhu/autogen/notebook/coding/tmp_code_0b52d7fcfe06d76c8b1e0ce9333cb5cf.py:14: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]` meta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100 /Users/ekzhu/autogen/notebook/coding/tmp_code_0b52d7fcfe06d76c8b1e0ce9333cb5cf.py:15: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]` tesla_ytd_gain = ((tesla_data['Close'][-1] - tesla_data['Close'][0]) / tesla_data['Close'][0]) * 100 META year-to-date gain: 50.11406747602124% TESLA year-to-date gain: -30.85903076529873% -------------------------------------------------------------------------------- assistant (to user_proxy): The code has successfully fetched the stock prices for META and TESLA and calculated the year-to-date gain for each. As of yesterday, the year-to-date gain for META (Meta Platforms Inc.) is approximately 50.11%, and for TESLA (Tesla Inc.) it is approximately -30.86%. This means that so far this year, META's stock price has increased by about 50.11% while TESLA's stock price has decreased by about 30.86%. Please note that stock prices can fluctuate and the exact gain may vary depending on the time of checking. TERMINATE --------------------------------------------------------------------------------
The example above involves code execution. In AutoGen, code execution is triggered automatically by the UserProxyAgent when it detects an executable code block in a received message and no human user input is provided.
Users have the option to specify a different working directory by setting the work_dir argument when constructing a new instance of the LocalCommandLineCodeExecutor.
For Docker-based or Jupyter kernel-based code execution, please refer to
Code Executors Tutorial for more information.
Check chat results¶
The initiate_chat method returns a ChatResult object, which is a dataclass object storing information about the chat. Currently, it includes the following attributes:
chat_history: a list of chat history.summary: a string of chat summary. A summary is only available if a summary_method is provided when initiating the chat.cost: a tuple of (total_cost, total_actual_cost), where total_cost is a dictionary of cost information, and total_actual_cost is a dictionary of information on the actual incurred cost with cache.human_input: a list of strings of human inputs solicited during the chat. (Note that since we are settinghuman_input_modetoNEVERin this notebook, this list is always empty.)
In [4]:
print("Chat history:", chat_res.chat_history)
print("Summary:", chat_res.summary)
print("Cost info:", chat_res.cost)
Chat history: [{'content': 'What date is today? Compare the year-to-date gain for META and TESLA.', 'role': 'assistant'}, {'content': 'First, let\'s get the current date using Python. \n\n```python\n# Python code\nfrom datetime import date\n\n# Get today\'s date\ntoday = date.today()\n\n# Print today\'s date\nprint("Today\'s date:", today)\n```\n\nNext, we need to fetch the stock prices for META (Facebook) and TESLA for the current year. We can use the `yfinance` library in Python to fetch this data. If `yfinance` is not installed, it can be installed using pip: `pip install yfinance`.\n\nHere is the Python code to fetch the stock prices and calculate the year-to-date gain:\n\n```python\n# Python code\nimport yfinance as yf\nfrom datetime import datetime\n\n# Get the current year\ncurrent_year = datetime.now().year\n\n# Download stock data for the current year\nmeta_data = yf.download(\'FB\', start=f\'{current_year}-01-01\', end=today)\ntesla_data = yf.download(\'TSLA\', start=f\'{current_year}-01-01\', end=today)\n\n# Calculate the year-to-date gain for each stock\nmeta_ytd_gain = ((meta_data[\'Close\'][-1] - meta_data[\'Close\'][0]) / meta_data[\'Close\'][0]) * 100\ntesla_ytd_gain = ((tesla_data[\'Close\'][-1] - tesla_data[\'Close\'][0]) / tesla_data[\'Close\'][0]) * 100\n\n# Print the year-to-date gain for each stock\nprint(f\'META year-to-date gain: {meta_ytd_gain}%\')\nprint(f\'TESLA year-to-date gain: {tesla_ytd_gain}%\')\n```\n\nThis code fetches the closing prices for META and TESLA for the current year, calculates the year-to-date gain for each stock, and prints the results. The year-to-date gain is calculated as the percentage change in the closing price from the first trading day of the year to the most recent trading day.', 'role': 'user'}, {'content': 'exitcode: 1 (execution failed)\nCode output: Today\'s date: 2024-04-12\nTraceback (most recent call last):\n File "/Users/ekzhu/autogen/notebook/coding/tmp_code_cb9ef30baa23cf28e127198c0ebeb7e6.py", line 9, in <module>\n meta_data = yf.download(\'FB\', start=f\'{current_year}-01-01\', end=today)\n ^^^^^\nNameError: name \'today\' is not defined\n', 'role': 'assistant'}, {'content': "I apologize for the oversight. The variable `today` was defined in the first code block but not in the second one. Let's correct this by defining `today` in the second code block as well. Here's the corrected code:\n\n```python\n# Python code\nimport yfinance as yf\nfrom datetime import datetime, date\n\n# Get the current year and today's date\ncurrent_year = datetime.now().year\ntoday = date.today()\n\n# Download stock data for the current year\nmeta_data = yf.download('FB', start=f'{current_year}-01-01', end=today)\ntesla_data = yf.download('TSLA', start=f'{current_year}-01-01', end=today)\n\n# Calculate the year-to-date gain for each stock\nmeta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100\ntesla_ytd_gain = ((tesla_data['Close'][-1] - tesla_data['Close'][0]) / tesla_data['Close'][0]) * 100\n\n# Print the year-to-date gain for each stock\nprint(f'META year-to-date gain: {meta_ytd_gain}%')\nprint(f'TESLA year-to-date gain: {tesla_ytd_gain}%')\n```\n\nThis code should now correctly fetch the stock prices for META and TESLA, calculate the year-to-date gain for each, and print the results.", 'role': 'user'}, {'content': 'exitcode: 1 (execution failed)\nCode output: \n[*********************100%%**********************] 1 of 1 completed\n\n1 Failed download:\n[\'FB\']: Exception(\'%ticker%: No timezone found, symbol may be delisted\')\n/Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/yfinance/utils.py:775: FutureWarning: The \'unit\' keyword in TimedeltaIndex construction is deprecated and will be removed in a future version. Use pd.to_timedelta instead.\n df.index += _pd.TimedeltaIndex(dst_error_hours, \'h\')\n\n[*********************100%%**********************] 1 of 1 completed\n/Users/ekzhu/autogen/notebook/coding/tmp_code_9ca584c3d8b484c70ee49c1ce5e4b975.py:14: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`\n meta_ytd_gain = ((meta_data[\'Close\'][-1] - meta_data[\'Close\'][0]) / meta_data[\'Close\'][0]) * 100\nTraceback (most recent call last):\n File "/Users/ekzhu/autogen/notebook/coding/tmp_code_9ca584c3d8b484c70ee49c1ce5e4b975.py", line 14, in <module>\n meta_ytd_gain = ((meta_data[\'Close\'][-1] - meta_data[\'Close\'][0]) / meta_data[\'Close\'][0]) * 100\n ~~~~~~~~~~~~~~~~~~^^^^\n File "/Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/pandas/core/series.py", line 1109, in __getitem__\n return self._values[key]\n ~~~~~~~~~~~~^^^^^\nIndexError: index -1 is out of bounds for axis 0 with size 0\n', 'role': 'assistant'}, {'content': "I apologize for the confusion. It seems like the ticker symbol for Meta Platforms Inc. (formerly Facebook) has changed from 'FB' to 'META'. Let's correct this in the code. Also, it seems like the data for the current day might not be available yet, so let's fetch the data until the previous day. Here's the corrected code:\n\n```python\n# Python code\nimport yfinance as yf\nfrom datetime import datetime, timedelta\n\n# Get the current year and yesterday's date\ncurrent_year = datetime.now().year\nyesterday = datetime.now() - timedelta(1)\n\n# Download stock data for the current year\nmeta_data = yf.download('META', start=f'{current_year}-01-01', end=yesterday)\ntesla_data = yf.download('TSLA', start=f'{current_year}-01-01', end=yesterday)\n\n# Calculate the year-to-date gain for each stock\nmeta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100\ntesla_ytd_gain = ((tesla_data['Close'][-1] - tesla_data['Close'][0]) / tesla_data['Close'][0]) * 100\n\n# Print the year-to-date gain for each stock\nprint(f'META year-to-date gain: {meta_ytd_gain}%')\nprint(f'TESLA year-to-date gain: {tesla_ytd_gain}%')\n```\n\nThis code should now correctly fetch the stock prices for META and TESLA, calculate the year-to-date gain for each, and print the results.", 'role': 'user'}, {'content': "exitcode: 0 (execution succeeded)\nCode output: /Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/yfinance/utils.py:775: FutureWarning: The 'unit' keyword in TimedeltaIndex construction is deprecated and will be removed in a future version. Use pd.to_timedelta instead.\n df.index += _pd.TimedeltaIndex(dst_error_hours, 'h')\n\n[*********************100%%**********************] 1 of 1 completed\n/Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/yfinance/utils.py:775: FutureWarning: The 'unit' keyword in TimedeltaIndex construction is deprecated and will be removed in a future version. Use pd.to_timedelta instead.\n df.index += _pd.TimedeltaIndex(dst_error_hours, 'h')\n\n[*********************100%%**********************] 1 of 1 completed\n/Users/ekzhu/autogen/notebook/coding/tmp_code_0b52d7fcfe06d76c8b1e0ce9333cb5cf.py:14: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`\n meta_ytd_gain = ((meta_data['Close'][-1] - meta_data['Close'][0]) / meta_data['Close'][0]) * 100\n/Users/ekzhu/autogen/notebook/coding/tmp_code_0b52d7fcfe06d76c8b1e0ce9333cb5cf.py:15: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`\n tesla_ytd_gain = ((tesla_data['Close'][-1] - tesla_data['Close'][0]) / tesla_data['Close'][0]) * 100\nMETA year-to-date gain: 50.11406747602124%\nTESLA year-to-date gain: -30.85903076529873%\n", 'role': 'assistant'}, {'content': "The code has successfully fetched the stock prices for META and TESLA and calculated the year-to-date gain for each. \n\nAs of yesterday, the year-to-date gain for META (Meta Platforms Inc.) is approximately 50.11%, and for TESLA (Tesla Inc.) it is approximately -30.86%. This means that so far this year, META's stock price has increased by about 50.11% while TESLA's stock price has decreased by about 30.86%.\n\nPlease note that stock prices can fluctuate and the exact gain may vary depending on the time of checking.\n\nTERMINATE", 'role': 'user'}]
Summary: The year-to-date gain for META (Meta Platforms Inc.) is approximately 50.11%, and for TESLA (Tesla Inc.) it is approximately -30.86%. This means that so far this year, META's stock price has increased by about 50.11% while TESLA's stock price has decreased by about 30.86%.
Cost info: ({'total_cost': 0.32256, 'gpt-4-0613': {'cost': 0.32256, 'prompt_tokens': 8224, 'completion_tokens': 1264, 'total_tokens': 9488}}, {'total_cost': 0.32256, 'gpt-4-0613': {'cost': 0.32256, 'prompt_tokens': 8224, 'completion_tokens': 1264, 'total_tokens': 9488}})
Example Task: Plot Chart¶
In [5]:
# followup of the previous question
user_proxy.send(
recipient=assistant,
message="""Plot a chart of their stock price change YTD. Save the data to stock_price_ytd.csv, and save the plot to stock_price_ytd.png.""",
)
user_proxy (to assistant): Plot a chart of their stock price change YTD. Save the data to stock_price_ytd.csv, and save the plot to stock_price_ytd.png. -------------------------------------------------------------------------------- assistant (to user_proxy): To plot a chart of the stock price change year-to-date (YTD) for META and TESLA, and to save the data and the plot, we can use the `matplotlib` and `pandas` libraries in Python. If these libraries are not installed, they can be installed using pip: `pip install matplotlib pandas`. Here is the Python code to fetch the stock prices, plot the chart, and save the data and the plot: ```python # Python code import yfinance as yf import matplotlib.pyplot as plt import pandas as pd from datetime import datetime, timedelta # Get the current year and yesterday's date current_year = datetime.now().year yesterday = datetime.now() - timedelta(1) # Download stock data for the current year meta_data = yf.download('META', start=f'{current_year}-01-01', end=yesterday) tesla_data = yf.download('TSLA', start=f'{current_year}-01-01', end=yesterday) # Plot the closing prices plt.figure(figsize=(14, 7)) plt.plot(meta_data['Close'], label='META') plt.plot(tesla_data['Close'], label='TESLA') plt.title('META vs TESLA Stock Price YTD') plt.xlabel('Date') plt.ylabel('Closing Price') plt.legend() plt.grid(True) plt.savefig('stock_price_ytd.png') # Save the data to a CSV file data = pd.concat([meta_data['Close'], tesla_data['Close']], axis=1) data.columns = ['META', 'TESLA'] data.to_csv('stock_price_ytd.csv') ``` This code fetches the closing prices for META and TESLA for the current year, plots the closing prices, saves the plot to 'stock_price_ytd.png', and saves the closing prices to 'stock_price_ytd.csv'. -------------------------------------------------------------------------------- >>>>>>>> EXECUTING CODE BLOCK (inferred language is python)... user_proxy (to assistant): exitcode: 0 (execution succeeded) Code output: /Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/yfinance/utils.py:775: FutureWarning: The 'unit' keyword in TimedeltaIndex construction is deprecated and will be removed in a future version. Use pd.to_timedelta instead. df.index += _pd.TimedeltaIndex(dst_error_hours, 'h') [*********************100%%**********************] 1 of 1 completed /Users/ekzhu/miniconda3/envs/autogen/lib/python3.11/site-packages/yfinance/utils.py:775: FutureWarning: The 'unit' keyword in TimedeltaIndex construction is deprecated and will be removed in a future version. Use pd.to_timedelta instead. df.index += _pd.TimedeltaIndex(dst_error_hours, 'h') [*********************100%%**********************] 1 of 1 completed -------------------------------------------------------------------------------- assistant (to user_proxy): The code has successfully fetched the stock prices for META and TESLA, plotted the chart of their stock price change year-to-date (YTD), and saved the data to 'stock_price_ytd.csv' and the plot to 'stock_price_ytd.png'. You can now view the chart in the 'stock_price_ytd.png' file and the data in the 'stock_price_ytd.csv' file in your current working directory. TERMINATE --------------------------------------------------------------------------------
Let's display the generated figure.
In [6]:
try:
image = Image(filename="coding/stock_price_ytd.png")
display(image)
except FileNotFoundError:
print("Image not found. Please check the file name and modify if necessary.")
Let's display the raw data collected and saved from previous chat as well.
In [7]:
# Path to your CSV file
file_path = "coding/stock_price_ytd.csv"
try:
with open(file_path, mode="r", encoding="utf-8") as file:
# Read each line in the file
for line in file:
# Split the line into a list using the comma as a separator
row = line.strip().split(",")
# Print the list representing the current row
print(row)
except FileNotFoundError:
print("File not found. Please check the file name and modify if necessary.")
['Date', 'META', 'TESLA'] ['2024-01-02', '346.2900085449219', '248.4199981689453'] ['2024-01-03', '344.4700012207031', '238.4499969482422'] ['2024-01-04', '347.1199951171875', '237.92999267578125'] ['2024-01-05', '351.95001220703125', '237.49000549316406'] ['2024-01-08', '358.6600036621094', '240.4499969482422'] ['2024-01-09', '357.42999267578125', '234.9600067138672'] ['2024-01-10', '370.4700012207031', '233.94000244140625'] ['2024-01-11', '369.6700134277344', '227.22000122070312'] ['2024-01-12', '374.489990234375', '218.88999938964844'] ['2024-01-16', '367.4599914550781', '219.91000366210938'] ['2024-01-17', '368.3699951171875', '215.5500030517578'] ['2024-01-18', '376.1300048828125', '211.8800048828125'] ['2024-01-19', '383.45001220703125', '212.19000244140625'] ['2024-01-22', '381.7799987792969', '208.8000030517578'] ['2024-01-23', '385.20001220703125', '209.13999938964844'] ['2024-01-24', '390.70001220703125', '207.8300018310547'] ['2024-01-25', '393.17999267578125', '182.6300048828125'] ['2024-01-26', '394.1400146484375', '183.25'] ['2024-01-29', '401.0199890136719', '190.92999267578125'] ['2024-01-30', '400.05999755859375', '191.58999633789062'] ['2024-01-31', '390.1400146484375', '187.2899932861328'] ['2024-02-01', '394.7799987792969', '188.86000061035156'] ['2024-02-02', '474.989990234375', '187.91000366210938'] ['2024-02-05', '459.4100036621094', '181.05999755859375'] ['2024-02-06', '454.7200012207031', '185.10000610351562'] ['2024-02-07', '469.5899963378906', '187.5800018310547'] ['2024-02-08', '470.0', '189.55999755859375'] ['2024-02-09', '468.1099853515625', '193.57000732421875'] ['2024-02-12', '468.8999938964844', '188.1300048828125'] ['2024-02-13', '460.1199951171875', '184.02000427246094'] ['2024-02-14', '473.2799987792969', '188.7100067138672'] ['2024-02-15', '484.0299987792969', '200.4499969482422'] ['2024-02-16', '473.32000732421875', '199.9499969482422'] ['2024-02-20', '471.75', '193.75999450683594'] ['2024-02-21', '468.0299987792969', '194.77000427246094'] ['2024-02-22', '486.1300048828125', '197.41000366210938'] ['2024-02-23', '484.0299987792969', '191.97000122070312'] ['2024-02-26', '481.739990234375', '199.39999389648438'] ['2024-02-27', '487.04998779296875', '199.72999572753906'] ['2024-02-28', '484.0199890136719', '202.0399932861328'] ['2024-02-29', '490.1300048828125', '201.8800048828125'] ['2024-03-01', '502.29998779296875', '202.63999938964844'] ['2024-03-04', '498.19000244140625', '188.13999938964844'] ['2024-03-05', '490.2200012207031', '180.74000549316406'] ['2024-03-06', '496.0899963378906', '176.5399932861328'] ['2024-03-07', '512.1900024414062', '178.64999389648438'] ['2024-03-08', '505.95001220703125', '175.33999633789062'] ['2024-03-11', '483.5899963378906', '177.77000427246094'] ['2024-03-12', '499.75', '177.5399932861328'] ['2024-03-13', '495.57000732421875', '169.47999572753906'] ['2024-03-14', '491.8299865722656', '162.5'] ['2024-03-15', '484.1000061035156', '163.57000732421875'] ['2024-03-18', '496.9800109863281', '173.8000030517578'] ['2024-03-19', '496.239990234375', '171.32000732421875'] ['2024-03-20', '505.5199890136719', '175.66000366210938'] ['2024-03-21', '507.760009765625', '172.82000732421875'] ['2024-03-22', '509.5799865722656', '170.8300018310547'] ['2024-03-25', '503.0199890136719', '172.6300048828125'] ['2024-03-26', '495.8900146484375', '177.6699981689453'] ['2024-03-27', '493.8599853515625', '179.8300018310547'] ['2024-03-28', '485.5799865722656', '175.7899932861328'] ['2024-04-01', '491.3500061035156', '175.22000122070312'] ['2024-04-02', '497.3699951171875', '166.6300048828125'] ['2024-04-03', '506.739990234375', '168.3800048828125'] ['2024-04-04', '510.9200134277344', '171.11000061035156'] ['2024-04-05', '527.3400268554688', '164.89999389648438'] ['2024-04-08', '519.25', '172.97999572753906'] ['2024-04-09', '516.9000244140625', '176.8800048828125'] ['2024-04-10', '519.8300170898438', '171.75999450683594']
Example Task: Use User Defined Message Function to let Agents Analyze data Collected¶
Let's create a user defined message to let the agents analyze the raw data and write a blogpost. The function is supposed to take sender, recipient and context as inputs and outputs a string of message.
**kwargs from initiate_chat will be used as context. Take the following code as an example, the context includes a field file_name as provided in initiate_chat. In the user defined message function my_message_generator, we are reading data from the file specified by this filename.
In [11]:
def my_message_generator(sender, recipient, context):
# your CSV file
file_name = context.get("file_name")
try:
with open(file_name, mode="r", encoding="utf-8") as file:
file_content = file.read()
except FileNotFoundError:
file_content = "No data found."
return "Analyze the data and write a brief but engaging blog post. \n Data: \n" + file_content
# followup of the previous question
chat_res = user_proxy.initiate_chat(
recipient=assistant,
message=my_message_generator,
file_name="coding/stock_price_ytd.csv",
summary_method="reflection_with_llm",
summary_args={"summary_prompt": "Return the blog post in Markdown format."},
)
user_proxy (to assistant): Analyze the data and write a brief but engaging blog post. Data: Date,META,TESLA 2024-01-02,346.2900085449219,248.4199981689453 2024-01-03,344.4700012207031,238.4499969482422 2024-01-04,347.1199951171875,237.92999267578125 2024-01-05,351.95001220703125,237.49000549316406 2024-01-08,358.6600036621094,240.4499969482422 2024-01-09,357.42999267578125,234.9600067138672 2024-01-10,370.4700012207031,233.94000244140625 2024-01-11,369.6700134277344,227.22000122070312 2024-01-12,374.489990234375,218.88999938964844 2024-01-16,367.4599914550781,219.91000366210938 2024-01-17,368.3699951171875,215.5500030517578 2024-01-18,376.1300048828125,211.8800048828125 2024-01-19,383.45001220703125,212.19000244140625 2024-01-22,381.7799987792969,208.8000030517578 2024-01-23,385.20001220703125,209.13999938964844 2024-01-24,390.70001220703125,207.8300018310547 2024-01-25,393.17999267578125,182.6300048828125 2024-01-26,394.1400146484375,183.25 2024-01-29,401.0199890136719,190.92999267578125 2024-01-30,400.05999755859375,191.58999633789062 2024-01-31,390.1400146484375,187.2899932861328 2024-02-01,394.7799987792969,188.86000061035156 2024-02-02,474.989990234375,187.91000366210938 2024-02-05,459.4100036621094,181.05999755859375 2024-02-06,454.7200012207031,185.10000610351562 2024-02-07,469.5899963378906,187.5800018310547 2024-02-08,470.0,189.55999755859375 2024-02-09,468.1099853515625,193.57000732421875 2024-02-12,468.8999938964844,188.1300048828125 2024-02-13,460.1199951171875,184.02000427246094 2024-02-14,473.2799987792969,188.7100067138672 2024-02-15,484.0299987792969,200.4499969482422 2024-02-16,473.32000732421875,199.9499969482422 2024-02-20,471.75,193.75999450683594 2024-02-21,468.0299987792969,194.77000427246094 2024-02-22,486.1300048828125,197.41000366210938 2024-02-23,484.0299987792969,191.97000122070312 2024-02-26,481.739990234375,199.39999389648438 2024-02-27,487.04998779296875,199.72999572753906 2024-02-28,484.0199890136719,202.0399932861328 2024-02-29,490.1300048828125,201.8800048828125 2024-03-01,502.29998779296875,202.63999938964844 2024-03-04,498.19000244140625,188.13999938964844 2024-03-05,490.2200012207031,180.74000549316406 2024-03-06,496.0899963378906,176.5399932861328 2024-03-07,512.1900024414062,178.64999389648438 2024-03-08,505.95001220703125,175.33999633789062 2024-03-11,483.5899963378906,177.77000427246094 2024-03-12,499.75,177.5399932861328 2024-03-13,495.57000732421875,169.47999572753906 2024-03-14,491.8299865722656,162.5 2024-03-15,484.1000061035156,163.57000732421875 2024-03-18,496.9800109863281,173.8000030517578 2024-03-19,496.239990234375,171.32000732421875 2024-03-20,505.5199890136719,175.66000366210938 2024-03-21,507.760009765625,172.82000732421875 2024-03-22,509.5799865722656,170.8300018310547 2024-03-25,503.0199890136719,172.6300048828125 2024-03-26,495.8900146484375,177.6699981689453 2024-03-27,493.8599853515625,179.8300018310547 2024-03-28,485.5799865722656,175.7899932861328 2024-04-01,491.3500061035156,175.22000122070312 2024-04-02,497.3699951171875,166.6300048828125 2024-04-03,506.739990234375,168.3800048828125 2024-04-04,510.9200134277344,171.11000061035156 2024-04-05,527.3400268554688,164.89999389648438 2024-04-08,519.25,172.97999572753906 2024-04-09,516.9000244140625,176.8800048828125 2024-04-10,519.8300170898438,171.75999450683594 -------------------------------------------------------------------------------- assistant (to user_proxy): To analyze this data and write a blog post, we first need to load the data into a pandas DataFrame, then perform some basic analysis on it. We can then use the results of this analysis to write the blog post. Here is the Python code to load the data and perform the analysis: ```python # Python code import pandas as pd from io import StringIO data = """ Date,META,TESLA 2024-01-02,346.2900085449219,248.4199981689453 2024-01-03,344.4700012207031,238.4499969482422 ... 2024-04-10,519.8300170898438,171.75999450683594 """ # Load the data into a pandas DataFrame df = pd.read_csv(StringIO(data)) # Calculate the mean, max, min, and standard deviation for each stock meta_mean = df['META'].mean() meta_max = df['META'].max() meta_min = df['META'].min() meta_std = df['META'].std() tesla_mean = df['TESLA'].mean() tesla_max = df['TESLA'].max() tesla_min = df['TESLA'].min() tesla_std = df['TESLA'].std() # Print the results print(f"META - Mean: {meta_mean}, Max: {meta_max}, Min: {meta_min}, Std: {meta_std}") print(f"TESLA - Mean: {tesla_mean}, Max: {tesla_max}, Min: {tesla_min}, Std: {tesla_std}") ``` After running this code, we can use the results to write the blog post. Here is a possible blog post based on the results: --- Title: A Comparative Analysis of META and TESLA Stocks in Early 2024 In the first quarter of 2024, the stock market saw some interesting movements in the tech sector. Two companies that stood out during this period were META and TESLA. META, the social media giant, had an average stock price of [META Mean] during this period. The highest it reached was [META Max], while the lowest was [META Min]. The standard deviation, a measure of how spread out the prices were, was [META Std]. On the other hand, TESLA, the electric vehicle and clean energy company, had an average stock price of [TESLA Mean]. The stock reached a high of [TESLA Max] and a low of [TESLA Min]. The standard deviation for TESLA was [TESLA Std]. These figures show that both META and TESLA had their ups and downs during this period. However, the higher standard deviation for [Company with higher Std] indicates that its stock price fluctuated more compared to [Company with lower Std]. As we move further into 2024, it will be interesting to see how these trends evolve. Will META and TESLA continue on their current trajectories, or will we see a shift in the market dynamics? Only time will tell. --- Please replace [META Mean], [META Max], [META Min], [META Std], [TESLA Mean], [TESLA Max], [TESLA Min], [TESLA Std], [Company with higher Std], and [Company with lower Std] with the actual values obtained from the Python code. -------------------------------------------------------------------------------- >>>>>>>> EXECUTING CODE BLOCK (inferred language is python)... user_proxy (to assistant): exitcode: 0 (execution succeeded) Code output: META - Mean: 403.53000895182294, Max: 519.8300170898438, Min: 344.4700012207031, Std: 100.72287240911488 TESLA - Mean: 219.54332987467447, Max: 248.4199981689453, Min: 171.75999450683594, Std: 41.68075797345465 -------------------------------------------------------------------------------- assistant (to user_proxy): Great! Now that we have the results, let's write the blog post. --- Title: A Comparative Analysis of META and TESLA Stocks in Early 2024 In the first quarter of 2024, the stock market saw some interesting movements in the tech sector. Two companies that stood out during this period were META and TESLA. META, the social media giant, had an average stock price of 403.53 during this period. The highest it reached was 519.83, while the lowest was 344.47. The standard deviation, a measure of how spread out the prices were, was 100.72. On the other hand, TESLA, the electric vehicle and clean energy company, had an average stock price of 219.54. The stock reached a high of 248.42 and a low of 171.76. The standard deviation for TESLA was 41.68. These figures show that both META and TESLA had their ups and downs during this period. However, the higher standard deviation for META indicates that its stock price fluctuated more compared to TESLA. As we move further into 2024, it will be interesting to see how these trends evolve. Will META and TESLA continue on their current trajectories, or will we see a shift in the market dynamics? Only time will tell. --- This blog post provides a brief but engaging analysis of the META and TESLA stocks based on the provided data. It highlights the mean, max, min, and standard deviation of the stock prices for both companies, and provides some commentary on what these figures might mean for the future. -------------------------------------------------------------------------------- user_proxy (to assistant): -------------------------------------------------------------------------------- assistant (to user_proxy): TERMINATE --------------------------------------------------------------------------------
Let's check the summary of the chat
In [12]:
print(chat_res.summary)
--- # A Comparative Analysis of META and TESLA Stocks in Early 2024 In the first quarter of 2024, the stock market saw some interesting movements in the tech sector. Two companies that stood out during this period were META and TESLA. META, the social media giant, had an average stock price of 403.53 during this period. The highest it reached was 519.83, while the lowest was 344.47. The standard deviation, a measure of how spread out the prices were, was 100.72. On the other hand, TESLA, the electric vehicle and clean energy company, had an average stock price of 219.54. The stock reached a high of 248.42 and a low of 171.76. The standard deviation for TESLA was 41.68. These figures show that both META and TESLA had their ups and downs during this period. However, the higher standard deviation for META indicates that its stock price fluctuated more compared to TESLA. As we move further into 2024, it will be interesting to see how these trends evolve. Will META and TESLA continue on their current trajectories, or will we see a shift in the market dynamics? Only time will tell. ---
This is the blog post that the agents generated.
A Comparative Analysis of META and TESLA Stocks in Early 2024¶
In the first quarter of 2024, the stock market saw some interesting movements in the tech sector. Two companies that stood out during this period were META and TESLA.
META, the social media giant, had an average stock price of 403.53 during this period. The highest it reached was 519.83, while the lowest was 344.47. The standard deviation, a measure of how spread out the prices were, was 100.72.
On the other hand, TESLA, the electric vehicle and clean energy company, had an average stock price of 219.54. The stock reached a high of 248.42 and a low of 171.76. The standard deviation for TESLA was 41.68.
These figures show that both META and TESLA had their ups and downs during this period. However, the higher standard deviation for META indicates that its stock price fluctuated more compared to TESLA.
As we move further into 2024, it will be interesting to see how these trends evolve. Will META and TESLA continue on their current trajectories, or will we see a shift in the market dynamics? Only time will tell.
Let's check how much the above chat cost
In [10]:
print(chat_res.cost)
({'total_cost': 0.84249, 'gpt-4-0613': {'cost': 0.84249, 'prompt_tokens': 22513, 'completion_tokens': 2785, 'total_tokens': 25298}}, {'total_cost': 0.84249, 'gpt-4-0613': {'cost': 0.84249, 'prompt_tokens': 22513, 'completion_tokens': 2785, 'total_tokens': 25298}})