Make predictions based on historical data CSV file?

[marcelooliveira]

I'd appreciate some advice on a problem, since I don't have expertise in AI or Data Science.

I need to analyze historical data to build a forecasting model, using a CSV data set containing historical data. Then I need to predict future values based on a future date input by the user, using the historical time series present in the dataset.

I'd like to know the model that best suits my training. Does it make sense to base my training on LLama2 or Llama2 chat for this? If not, what's the best model to build upon? And how should I format my historical CSV file?

Thanks in advance!

[jeffxtang]

Depending on the characteristics of your historical data, traditional ML models such as regressions, KNN and decision trees (you can use scikit-learn to quickly train and test with your data), or neural network models such as RNN and LSTM may be used. Other specialized time series forecasting models such as Prophet can also be used.

LLMs such as Llama 2 may also be used for the forecasting - e.g. see this repo.

[marcelooliveira]

Hi @jeffxtang, that's nice, I'll look into llmtime repo, Thanks so much for the comprehensive answer!

[wukaixingxp]

Hi! It seems that a solution has been provided to the issue and there has not been a follow-up conversation for a long time. I will close this issue for now and feel free to reopen it if you have any questions!