Can we get more learner for regression like PLSRegression and Cubist in R/Python

Currently I'm working on project that use regression to predict value from VIR NIR Wave. The initial model were built using R and Python with Cubist and PLSRegression algorthm and then I tried to convert the code to c# with ML.NET. I use AutoML API to get perfect model for better accuracy. I'm using R2 parameter to compare the model accuracy. But I cannot beat the accuracy in the previous model which is built with Python/R. here I attached the result of my experiment with jupyter + ML.NET
[SoilSensingKitModelv2.zip](https://github.com/dotnet/machinelearning-samples/files/3746136/SoilSensingKitModelv2.zip)