# What Dimensionality Reduction method I can follow on a dataset that has Physics Parameter?

I am trying to model data related to Locomotive Train. We have a various set of parameters and we have the possibility to generate a few more parameters from this. Our model is currently using a lot of parameters due to which it takes more time during computation. Since we have a lot of feature parameters and there is a good level of importance each parameter carries on the final parameter, I would like to know the best dimensionality reduction methods that are suitable for this kind of data which has physics parameters.

• Could you list the variables you already have? Do you know which products of powers of the variables are dimensionless? – J.G. May 21 at 20:12
• I have variables like acceleration, horsepower, weight, number of the CAR, the weight of each car and so on. – Karthik Ravi May 25 at 22:39
• Your variables (and their dimensions) include acceleration ($LT^{-2}$), power ($L^2MT^{-3}$), and the total mass ($M$). We can multiply the latter by some multiple of gravity, another acceleration, to get the frictional force the engine's power must cancel out, viz. $P_\text{friction-canceller}=F_\text{friction}v$ for $v$ the train's speed. What do you want to estimate with dimensional analysis? – J.G. May 26 at 13:49