I have a dataset that contains 500 field observations. The data dimension is 500 by 9 (that means 500 rows and 9 columns/variables). Among them, 8 variables are independent, and one is dependent variable.
I used the best hyperparameters combination by grid search with cross-validation =5.
I am trying to train the following machine learning models: decision-tree, RF, and SVC; but I cannot improve accuracy above 60%.
I have a constraint: I cannot improve field observation;
In this circumstance, I seek expert opinions to improve my model accuracy.