Data Analysis with PandasAI and Ollama

No video

Data Analysis with PandasAI and Ollama - Locally and Free

Рет қаралды 7,979

Күн бұрын

Today, we'll cover how to perform data analysis with PandasAI and Ollama using Python. At the end of the video, with generative AI, you'll learn data analysis projects locally and for free.
🔗 Let's connect:
Medium: / tirendazacademy
X: x.com/tirendaz...
LinkedIn: / tirendaz-academy
00:01 Introduction
00:40 What is PandasAI?
01:05 Setup
02:55 Load the dataset
03:54 Initialize the model
06:10 Chat with your data
🔗 Project files: github.com/Tir...
▶️ LangChain Tutorials:
• LangChain Tutorials
▶️ Generative AI Tutorials:
• Generative AI Tutorials
▶️ LLMs Tutorials:
• LLMs Tutorials
▶️ HuggingFace Tutorials:
• HuggingFace Tutorials ...
🔥 Thanks for watching. Don't forget to subscribe, like the video, and leave a comment.
#ai #openai #generativeai

Пікірлер: 49

@HistoryIsAbsurd 5 ай бұрын

Definitely worth the sub! Was lookin for exactly this Edit: also, thank you!

@TirendazAI 5 ай бұрын

You're welcome!

@mapledev9335 4 ай бұрын

PandasAI is brilliant. Thanks for putting this video together!

@mapledev9335 4 ай бұрын

I used the same question but only get the table. Change it a bit and I get a bargraph. Rephrase the question again and I get an error. Any suggestions how to get both the table and chart like in your video?

@TirendazAI 4 ай бұрын

Prompt engineering is important to get good results. I got this response by trying a few prompts. You may write more detailed prompts.

@GhostCoder83 4 ай бұрын

Thanks Bro. Very Helpful.

@TirendazAI 4 ай бұрын

Glad it helped

@AliAlias 5 ай бұрын

Thanks very much ❤❤❤ I wait this video for long time 😊 Q: what is best open source LLM from hugging face used for PandasAi & SQL data analysis

@TirendazAI 5 ай бұрын

You're welcome. Choosing the best model varies from task to task. I love working with Mixtral for PandasAI.

@aroonsubway2079 4 ай бұрын

Thanks for this great video! One question from me: Since the classical pandas style data analysis performed well, what is the potential advantage of using LLM? Is it possible that LLM can introduce more intelligent analysis than excel?

@TirendazAI 4 ай бұрын

The advantage of this tool is that you can use it to explore, clean and analyze your data using generative artificial intelligence. All you need to do is talk to your data. If you put in good prompts, you can get good outputs. You can also use this library to analyze your Excel data.

@aroonsubway2079 4 ай бұрын

@@TirendazAI Yes, an user-friendly QA is definitely something important. Thanks! But do you think we can find a scenario that Excel fails because it is not intelligent enough, but LLM has a chance to outperform Excel?

@TirendazAI 4 ай бұрын

LLMs have huge potential for data analysis. Yes, they are not intelligent enough. We can think of them as a black box but, you can provide that they return good outputs using agents or RAG techniques.

@PuffNSnort 3 ай бұрын

Great video! What are the dataset size limitations? I get an answer 30% of the time and errors the rest of the time.

@TirendazAI 3 ай бұрын

Large models like Llama-3:70b and GPT-4 respond better.

@charlesoni2787 4 ай бұрын

Nice. Thanks

@TirendazAI 4 ай бұрын

You're welcome!

@user-mv9ul9tz1c 3 ай бұрын

The tutorial video was truly amazing, with very clear subtitle translations. However, I'd like to know how to use PandasAI + Ollama in VS Code like you demonstrated. Also, how can I implement Ollama + Llama3 + LlamaIndex? What are the minimum computer specifications required for such usage? Additionally, how should I write the Python code to ensure that your consecutive queries are processed in the same thread? Best regards;

@TirendazAI 3 ай бұрын

For example, you need a minimum of 7B RAM to work with a 7B LLM version. Unfortunately, the responses generated by LLMs are not stable. You need to try several prompts to get the output you want.

@user-mv9ul9tz1c 3 ай бұрын

@@TirendazAI Thank you very much. However, my current computer setup includes an i7 processor, an RTX 2060 6G graphics card, and 16GB RAM. Can I run a local LLM with these specifications?

@TirendazAI 3 ай бұрын

Yes, you can run a local LLM such as Llama3:8b or mistral:7b.

@user-mv9ul9tz1c 3 ай бұрын

@@TirendazAI Thank you for your response. Originally, I planned to spend about $120 to upgrade my laptop's standard RAM from 16GB to 64GB. Additionally, could you tell me what software you used in your video demonstration? The interface looks very similar to VS Code, which I am currently trying to learn on my own. However, I'm not sure how to set up the environment. My command prompt shows Python 3.10.6. Do you have any videos that explain how to install Python and use related editors? Setting up the Python environment seems a bit complex, as it appears you have to set the environment before running the program.

@deepakkapoor5427 4 ай бұрын

facing this error 'Unfortunately, I was not able to answer your question, because of the following error: No code found in the response '

@TirendazAI 3 ай бұрын

When a prompt does not work, I am trying again by changing this prompt.

@sorgulabiraz3161 18 күн бұрын

Thank you. I've got an error like this: NameError Traceback (most recent call last) Cell In[30], line 2 1 from pandasai import SmartDataframe ----> 2 df = SmartDataframe(data, config={"llm": llm}) NameError: name 'data' is not defined

@Lifes_Student 12 күн бұрын

You had a traceback error and your first thought is to put it in a youtube comment rather then something like ChatGPT? ....Wow

@sorgulabiraz3161 12 күн бұрын

@@Lifes_Student I asked ChatGPT before and it didn't give me a solution. I am a beginner and I think I am free to ask any questions in any platform. Is there a problem for you? Who are you? Thougt police?

@Lifes_Student 12 күн бұрын

@@sorgulabiraz3161 lol ...I'm just surprised someone's willing to wait for a solution in a comment rather then instantly debug it using a language model these these days. You should probably also know that you'll need a powerful graphics card to run the mistral model locally. I recommend an RTX 3060

@Lifes_Student 12 күн бұрын

And just an FYI "data" is a variable. You need to define it first .i.e: data = yourdata then you can call 'data' or pass it through a function e.g. function(data)

@sorgulabiraz3161 12 күн бұрын

@@Lifes_Student Thank you.

@ajithnaidu6978 4 ай бұрын

Hi , Unfortunately I'm not able to install pandasai in the terminal it showing could not find a version that satisfies the requirement pandasai

@TirendazAI 4 ай бұрын

Hi, did you create a virtual environment? If yes, you can also use this command: poetry add pandasai

@ajithnaidu6978 4 ай бұрын

This poetry add pandasai used in the terminal or cell?

@sonidosdetranquilidad 5 ай бұрын

It does not work for me. I got the following error message: 40 if llm is None or not isinstance(llm, LLM or LangchainLLM): ---> 41 raise LLMNotFoundError("LLM is required") 42 return llm LLMNotFoundError: LLM is required

@TirendazAI 5 ай бұрын

Make sure you install Ollama and download a model

@JAlcocerTech 5 ай бұрын

Hello @tirendazakademi, Same Error for me. I have ollama version 0.1.28 Used: ollama pull mistral Tried with this dependencies (the ones that were latest 1 week ago when you uploaded) and got the very same error about: LLMNotFoundError("LLM is required") pandasai==2.0 langchain==0.1.10 langchain-community==0.0.25 Could you please verify which exact packages you used? Thanks and keep the great content!😁

@AlessandroPerugini-hz7pk 5 ай бұрын

i have the same error, langllm work with langchain, i can call : llm.invoke("Tell me a joke") with a valid response. but SmartDataFrame does not recognize ollama as a valid LLM. i have also tried to install the github package. The only difference is ollama version is 0.1.28

@AlessandroPerugini-hz7pk 5 ай бұрын

@@JAlcocerTech with miniconda i have your same version and error , with visual studio code seem to work these are the library version i have : pandas 1.5.3 pandasai 2.0.8 langchain 0.1.11 langchain-community 0.0.27 langchain-core 0.1.30 langchain-text-splitters 0.0.1 I hope this can help