Prophet Optimization with Particle Swarm Optimization (PSO) for Adjusted Closing Price Forecasting in Xiaomi-Corporation

Details :

1. Optimizing Prophet's hyperparameters, such as changepoint_prior_scale, seasonality_prior_scale, and holidays_prior_scale, with Particle Swarm Optimization (PSO)
2. Optimizing Prophet's forecasting accuracy by incorporating the Prophet model with Additional Regressors, such as Open, High, Low, Close, and Volume
3. This project uses Xiaomi Corporation Stock Prices data that was obtained from Yahoo Finance (https://finance.yahoo.com/quote/1810.HK/history)
4. The Prophet model was not given with a custom list of holidays that gives effect to the value of the Adjusted Closing Price of Xiaomi Corporation
5. The Prophet model uses the "Piecewise Linear" formula to calculate the Trend Component / g(t)
6. There are 4 models designed in this project, which are "Standard Prophet" / "Prophet", "Standard Prophet with Additional Regressors" / "Prophet + Reg", "Standard Prophet Optimized with PSO" / "Prophet + PSO", and "Standard Prophet with Additional Regressors and Optimized with PSO" / "Prophet + Reg + PSO"

Result :

The "Prophet + Reg + PSO" model achieved a high level of accuracy with MAPE and R2 values of 0.00000008522 and 0.999, respectively. This means the "Prophet + Reg + PSO" model is very accurate in forecasting the Adjusted Closing Prices of Xiaomi Corporation.