Evaluation of Machine Learning Model through Stock Price Prediction Research

This research holds paramount importance in advancing our utilization of artificial intelligence to predict economic factors, notably within the dynamic domain of the stock market. The primary objectives focus on determining the optimal performance among the seven machine learning models employed. Stock investment prices are never still; they are always changing. It is important to stay informed on the upward or downward trends of the market to make future investments. To accustom the machine learning (ML) predictor to the multitude of possibilities that could categorize stock patterns, 7 different ML models were trained on 1250 pieces of open stock market data dating to the last 5 years by assigning weight values to all the models based on their accuracy. The neural network ends up predicting the stock price with its given data at a mediocre level at best, with MSE averages of 29.93 and 26.85 respectively. Its highest weight, tesla, ends up with only 0.013% of the total weightage. Results showed that two of the ML models, specifically the Linear Regression and the Random Sample Consensus (RANSAC) Regressor models consistently outperformed the other 5 models, both ending up with the highest weight values of around 0.5 when predicting for Amazon, Apple, and Tesla. Therefore, the RANSAC and Linear Regression models are the best models to rely on when predicting open stock market prices using ML. Future endeavors must continue this trajectory by expanding model capacities, incorporating richer data sources, and embracing AI-driven advancements to propel stock market predictability into new realms.