how to improve accuracy?

[behi123]

I wrote this code to fit a symbolic regression model using PySRRegressor. However, the equations generated by the model do not achieve high accuracy. Here is the code I used:
X = data[["temp", "additive", "stress", "cycle"]].values
y = data["jnr"].values

scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

model = PySRRegressor(
niterations=500,
binary_operators=["+", "-", "*", "/", "^"],
unary_operators=["exp", "log", "sqrt"],
population_size=200,
populations=1000,
maxsize=12,
maxdepth=12,
progress=True,
loss="loss(x, y) = (x - y)^2",
model_selection="accuracy",
verbosity=1,
random_state=42,
)

model.fit(X_scaled, y)
Despite trying to set reasonable values for parameters like maxsize, maxdepth, niterations, and using loss(x, y) = (x - y)^2, the accuracy of the generated equations is still low. How can I improve the performance of the model? Should I adjust the parameters or try a different approach? Any suggestions would be greatly appreciated!

[MilesCranmer]

Check out the tuning page: https://ai.damtp.cam.ac.uk/pysr/tuning/. Also worth browsing other discussions.