ModuleNotFoundError: No module named 'pandasai'

[Elicherla01]

🐛 Describe the bug

I have intsalled pandasai, openai and strealit using

'''pip3 install pandasai openai streamlit'''

I am getting below error

'''ModuleNotFoundError: No module named 'pandasai''', when I execute

from pandasai import PandasAI from pandasai.llm.openai import OpenAI

Am i missing something?

[gventuri]

Hi @Elicherla01 thanks for reaching us out. Which version of python do you have installed? Also, please try pip show pandasai to be sure it's installed on your environment!

[Elicherla01]

@gventuri , I am using Python 3.10.9 and

[gventuri]

This is actually very weird. Are you running it within a venv or a docker container? If so, are you sure pandasai has been installed within the venv/container?

[Elicherla01]

This is actually very weird. Are you running it within a venv or a docker container? If so, are you sure pandasai has been installed within the venv/container?

I am using venv. I found it weird too. I have Conda installed. Do you think that is causing any issues? By the way it is working well in Jupyter notebook. But not working when i run this from the command prompt. I am using VSCode as IDE.

[mblazic]

Sorry, maybe I will go a little but of topic here, but the exactly the same issue I have here, but with the ansible module. @Elicherla01, on which OS you are working with pandasai module?

[gventuri]

@Elicherla01 @mblazic where are you executing the code from? I'm pretty sure that it's a different environment the one in which you run the code.

[mblazic]

In my case I'm executing the from the virtual environment, created with python3,9, the module is installed from the pip repository. I'm not using pandasai in my case, but the ansible module. The issues look similar so I think somehow the root problem is in the environment itself.

[baseerkhan]

Is this problem resolved? I am seeing similar issue in venv. Name: pandasai Version: 0.6.4 Summary: Pandas AI is a Python library that integrates generative artificial intelligence capabilities into Pandas, making dataframes conversational. Home-page: Author: Gabriele Venturi Author-email: License: MIT Location: /Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages Requires: astor, ipython, matplotlib, openai, pandas, python-dotenv Required-by:

[jonbiemond]

As Gabriele mentioned the most likely cause is that you are not calling import pandasai from the same environment it was installed in. If you're using venv see the docs on how to activate your environment before you run your script: https://docs.python.org/3/library/venv.html#how-venvs-work

If that doesn't solve it, please try to repeat the steps you used to install pandasai for numpy. If successful, please share more details about the steps you took to install pandasai so it can be resolved.

[mblazic]

I have solved my issue while using virtual environment, created with python3,9. The solution was to specify the python version which ansible should use, also with the same version the virtual environment was created. I have added vars line: ansible_python_interpreter: /opt/automation/bin/python3.9 in the ansible block, where the issue was triggered.

Example:

- name: Create and start monitoring services
  community.docker.docker_compose:
    services:
      - node-exporter
    state: present
  vars:
    ansible_python_interpreter: /opt/automation/bin/python3.9

[baseerkhan]

Ditto. Solved the issue by creating python 3.9 virtual environment. Thank you.

[gventuri]

@baseerkhan so it wasn't working on python 3.11 and it works on python 3.9?

[baseerkhan]

It wasn't working on python 3.8.

[gventuri]

Oh then it makes totally sense. Thanks a lot for reporting @baseerkhan!

While @Elicherla01 you were using python 3.10, right? Did you manage to solve the issue?

[robertsonjbrad]

Been having this same issue as well. Been going crazy creating different virtual environments running different versions of python. Have also been trying difference versions of pandasai. My most recent try is exactly the same as the Google Collab version except that it's python 3.9 and not 3.10.

Any pointers?

[noamsiegel]

Had the same issue, but realized that the issue stemmed from my having another pandasai.py file created. After deleting that, everything worked on my system (macOS M1) on VS Code. Here are my packages + Python:

numpy==1.25.2 pandas==2.0.0 polars==0.19.2 python==3.11.5

Let me know if this works for anyone on this thread -- would like to see if this is a bigger issue.

[sumeet-cresen]

Hi @gventuri ,

Any idea or solution to the above error ???

I am using numy -1.22 , python-3.9

Anyone in the loop , please let me know how to resolve the issue.

[jonbiemond]

Hi @sumeet-cresen, try changing your import statement on line 13 to:

from pandasai.llm import AzureOpenAI

If that still fails, please make sure you have the latest version of PandasAI installed.

[sumeet-cresen]

Hi @sumeet-cresen, try changing your import statement on line 13 to:

from pandasai.llm import AzureOpenAI

If that still fails, please make sure you have the latest version of PandasAI installed.

I am using the last pandasAI version @jonbiemond , but unable to retrive the package , Do you have any other suggestions

[jonbiemond]

@sumeet-cresen If the corrected import statement is still failing, it's possible pandasai isn't installed in your environment. Please double check using your package manager. pip show pandasai

[sumeet-cresen]

Hi @jonbiemond ,

Unable to resolve the above error , any idea on this ???

[vijayproxima]

looks like Module not found error for 'import pandasai from PandasAI' is quite common. Something is really serious about this as I do not get such eror for any of the other imports. I am trying this by creating a virtual env. my python version is 3.11.6 and on LINUX. I checked with 'pip show pandasai' and it is fine. Please help. THis is the exact error: ImportError: cannot import name 'PandasAI' from 'pandasai' (/home/user/pandasAI/pandasaienv/lib/python3.11/site-packages/pandasai/init.py) output of pip show pandasai: Name: pandasai Version: 1.5.2 Summary: PandasAI is a Python library that integrates generative artificial intelligence capabilities into Pandas, making dataframes conversational. Home-page: Author: Gabriele Venturi Author-email: License: MIT Location: /home/vijayproxima/pandasAI/pandasaienv/lib/python3.11/site-packages Requires: astor, duckdb, faker, ipython, matplotlib, openai, pandas, pillow, pydantic, python-dotenv, scipy, sqlalchemy Required-by:

[sumeet-cresen]

Hi @vijayproxima ,

please downgrade the version to 1.4.10

• uninstall the previous and install the above version.

[vijayproxima]

Many thanks, after downgrading the version, my problem got solved,. Warm Regards

Vijay Jadhav Proxima Systems Pvt. Ltd Pune - 411045

On Wed, Nov 22, 2023 at 12:33 PM sumeet-cresen @.***> wrote:

Hi @vijayproxima https://github.com/vijayproxima ,

please downgrade the version to 1.4.10

• uninstall the previous and install the above version.

[image: image] https://user-images.githubusercontent.com/127079133/284816759-29c122dc-ecfa-43c1-bf5f-8777c76218a3.png

— Reply to this email directly, view it on GitHub https://github.com/gventuri/pandas-ai/issues/286#issuecomment-1822216836, or unsubscribe https://github.com/notifications/unsubscribe-auth/AZEMGN7J2DTOEU3YKNTA4F3YFWPT5AVCNFSM6AAAAAAZGRGCNSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRSGIYTMOBTGY . You are receiving this because you were mentioned.Message ID: @.***>

[vijayproxima]

looks like Module not found error for 'import pandasai from PandasAI' is quite common. Something is really serious about this as I do not get such eror for any of the other imports. I am trying this by creating a virtual env. my python version is 3.11.6 and on LINUX. I checked with 'pip show pandasai' and it is fine. Please help. THis is the exact error: ImportError: cannot import name 'PandasAI' from 'pandasai' (/home/user/pandasAI/pandasaienv/lib/python3.11/site-packages/pandasai/init.py) output of pip show pandasai: Name: pandasai Version: 1.5.2 Summary: PandasAI is a Python library that integrates generative artificial intelligence capabilities into Pandas, making dataframes conversational. Home-page: Author: Gabriele Venturi Author-email: License: MIT Location: /home/vijayproxima/pandasAI/pandasaienv/lib/python3.11/site-packages Requires: astor, duckdb, faker, ipython, matplotlib, openai, pandas, pillow, pydantic, python-dotenv, scipy, sqlalchemy Required-by:

As you suggested, I downgraded the version from 1.5.2 to 1.4.10 and the problem got solved.

[sumeet-cresen]

Hi @vijayproxima ,

Can you able to generate graphs from pandasAI ???

[vijayproxima]

Hi @vijayproxima ,

Can you able to generate graphs from pandasAI ???

if you mean charts then Yes, you can generate various types of charts like pie, bar, heatmap.

[sumeet-cresen]

Hi @vijayproxima ,

i am unable to generate charts other than Histogram ,

error is shown below ,

were you able to generate charts ??

[vijayproxima]

Hi @vijayproxima ,

i am unable to generate charts other than Histogram ,

error is shown below ,

were you able to generate charts ??

Try re-running it as this error comes intermittently. I was able to generate pie, bar, heatmap and bubble. I hope you are using pandasai's SmartDataframe

[vijayproxima]

Hi @gventuri I am using PostgreSQLConnector to connect to a table which has a column that contains JSON data. When I try to query, I get this error: TypeError: unhashable type: 'dict' . please help. However I am able to query other tables that do NOT have JSON data in any of the columns. Secondly, if i have multiple tables, looks like we need to create a connector for every table that sounds bit weird. If we are going to have separate connectors for every table, how can we query to get output based on multiple tables [when there is a relationship between tables] Let me provide more details: One of the columns [machine_details] in the table has JSON data similar to : { “desktop info” : {

	“OS”: “my os”,
	“Keys”: “release”,
	“Model”:”ABC001”
},
"sys info": {
		"Event": 0,
		"Kernel": 0,
		"System": 6703
},

} I can query to get the specific details from JSON as follows: SELECT machine_details ->> 'desktop info',device_summary ->>'sys info' FROM machines . I think if we add this in the prompt, then the AI would understand how to deal with JSON data in table columns.