Forecasting Cocoon Silk Prices Using Machine Learning for Sericultural Planning

Silk farming is one of the leading subsectors of agricultural practices, especially among peasants growing cocoons silk crops, they are greatly concerned with the issue on the instability of the prices of cocoons silk. Forecasting such changes in price can be quite a complex task given that things like the environmental factors, farming practices, and changes in the market, among others cause such changes. Some of these factors are not easy to incorporate in the traditional forecasting tools hence resulting to either over or under estimation which is not favourable to the silk farmers. In the present paper, there was the employment of a machine learning algorithm of Random Forest Regressor to enhance the likelihood of the price prediction. It is trained using history price data from minimum price, the maximum price and the year. As a preliminary process, the data was prepared by removing or handling any null value and putting date into the appropriate format so that the data would be coherent enough for the training process. Upon the completion of the training process, the developed model displayed very high performances such as a very low RMSE of 493.49, an R-squared of 0.99 and a high prediction accuracy of 99.21%. These figures indicate that in a way, it is possible to account for the various interaction patterns that occur between these prices. It also can help make predictions more exact and also decision-making, which would make the market of silk more stable and enhance the economic steadiness of people who are engaged in silk farming. It can be enlarged in forms that include more factors that can be available such as climate or the current market status so as to make the model more useful to the farmer, trader and the policy maker.