Modelling the FPL data and determing AI use Case

[pratik-IDWT]

What's Next?

1. Make a simple query that brings together the data for these tables

2. Run the query and store into a modelled table or tables
3. Generate some trending analytics

What's the AI use case?

This need some thought but I can do one of a few things:

• Create a public facing web based tool, that lets you talk to the dataset and retrieve approved analytics through training and retrieval-augmented generation (RAG) principles
• Let AI analyse the data and show me trends

Originally posted by @pratik-IDWT in https://github.com/In-Data-We-Trust/IDWT-All-Things-SQL-And-Python-And-Code/issues/6#issuecomment-1731551061

[pratik-IDWT]

Some Links with connecting AI to a database to use as a showcase

When reading/researching/using AI within an organisation, I would lean towards using the RAG approach, which can be used to obtain approved/defined queries, to return approved information.

Some links to refer to when I get to this:

• https://medium.com/dataherald/how-to-langchain-sqlchain-c7342dd41614
• https://hbr.org/2023/07/how-to-train-generative-ai-using-your-companys-data

[pratik-IDWT]

Another link explaining how to train a LLP model for free using their tool

https://huggingface.co/blog/Llama2-for-non-engineers

However I won't use it, but take the learnings of inputs/outputs. I suspect they will use the information in the future to train their models and makea commercial product