<![CDATA[Figuring out diamond prices in the market is actually pretty tricky because the prices go up and down a lot and depend heavily on what the diamond looks like physically. The main problem here is that people often struggle to guess the right price manually, which can easily lead to losing money if they guess wrong. Through this study, we really want to test and directly compare two smart Machine Learning tools, which are the Random Forest and LightGBM algorithms, to see which one is actually better and more reliable at forecasting these prices without just blindly guessing. For how we did it, we used a huge dataset containing about 53,940 past diamond sales as study material. This data came complete with nine main features like carat weight, cut quality, color pigment, and physical size. Before doing the math with the computer, we quickly threw away any weird or impossible data rows to keep it clean. Then, we changed all the text descriptions into numbers so the program could read them easily. Next, we split the whole dataset up so that 80 percent was used to teach the program, and the remaining 20 percent was kept purely for testing how smart their guesses really were. The test results clearly showed that Random Forest did a much better job overall. It hit a super high R² accuracy score of 0.9835, and its guesses were only off by an average of $38.89 (MAE) and an RMSE of $77.24. This easily beat the LightGBM program, which only got an R² score of 0.9830, an MAE of $41.28, and an RMSE of $78.27. In short, this research proves that the team-work or ensemble system in Random Forest is way more reliable, stable, and accurate for predicting diamond prices in the market.]]>
