Precise evaluation of crop canopy characteristics requires optical remote sensing (RS) equipped with solid procedures. Despite much research on the subject, algorithm performance employing RS still needs improvement. In a short amount of time, five distinct algorithms partial-least-squares regression (PLSR), support vector regression (SVR), and random forest regression (RFR) with feature selection were used to evaluate the LAI and CWC of rice canopy data. Results from the visible to short-wave electromagnetic spectrum of canopy reflectance were input into the algorithms. Researchers used two sets of repeated field trials in the 2014–15 and 2015–16 rice growing seasons to gather data on leaf area index (LAI) (600) and Canopy water content (CWC) (480). The coefficient of determination (R2) was used to assess the efficiency of each method. Compared to other algorithms, PLSRLW outperformed them with R2 values of 0.77 for LAI and 0.66 for CWC. In addition, for every model, created a kernel concentration estimator of the root mean formed error values using a bootstrapping strategy. By giving calibration samples, which have the same canopy assembly as the test samples, more weight, the findings indicated that an appropriate method might be used to increase the forecast accuracy of LAI and CWC. Calibration of the model should be done using an entire season's worth of canopy spectral data since sub-setting it leads to high error values in the test dataset.