TabPFN-TS Based Accurate Price Forecasting Model to Farmers in Time-Series Data

The main aim of the project is to create a simple and proper prediction method for prices concerning crops based on a TabPFN-TS. Farmers can easily predict prices beforehand and can also make proper decisions concerning profit in the agricultural market. The historical data set regarding crops are taken from government sites like agromarket. This data is also pre-processed, meaning that gaps in data would be eliminated, and data would be made adequate for implementation. ARIMA model with 250 samples. TabPFN-TS model with 250 samples of time series data(1000-5000 rows) to predict the crop prices. The historical crop price as time series is used to feed the model to predict and display the results giving a 95% accuracy index and also provide a faster speed for the calculation process compared to the existing systems such as the ARIMA$(60-70 \%)$, Moving Average ($<65 \%$) and other statistical methods. By implementing the TabPFN-TS model setup as a basic setup, it will provide a possible option to get a right and precise prediction (>90%) and the significance$(\mathbf{p}<\mathbf{0. 0 0 1})$for the buying product even for the farmer with a less precise historic record for buying a product in these markets. The new system will ensure that crop cost estimates are accurate and that this will help the farmers to make informed decisions in the market for a profitable crop sales. The new system will make sustainable farming possible.