Change regression coefficient to percentage change I am running basic regression in R, and the numbers I am working with are quite high. when I run the regression I receive the coefficient in numbers change.  For instance, the dependent variable is "price" and the independent is "square meters" then I get a coefficient that is 50,427.120***. I was wondering if there is a way to change it so I get results in percentage change?
This is the code i am running:
reg.model1 <- Price2 ~ Ownership - 1 + Age +
  BRA + Bedrooms + Balcony + Lotsize
reg1 <- lm(formula = reg.model1, data = Data.apartment)

stargazer(reg1, type = "text")

 A: I think what you're asking for is what is the percent change in price for a 1 unit change in an independent variable. For this, you log-transform your dependent variable (price) by changing your formula to
reg.model1 <- log(Price2) ~ Ownership - 1 + Age + BRA + Bedrooms + Balcony + Lotsize
To interpret the coefficient, exponentiate it, subtract 1, and multiply it by 100. For example, if you run the regression and the coefficient for Age comes out as 0.03, then a 1 unit increase in Age increases the price by $ (e^{0.03}-1) \times 100 = 3.04$%  on average.
This link here explains it much better. Just be careful that log-transforming doesn't actually give a worse fit than before. Perhaps try using a quadratic model like  reg.model1 <- Price2 ~ Ownership - 1 + Age + BRA + Bedrooms + Balcony + Lotsize + I(Lotsize^2) and comparing the performance of the two.
A: A regression coefficient is the change in the outcome variable per unit change in a predictor variable.
The simplest way to reduce the magnitudes of all your regression coefficients would be to change the scale of your outcome variable. For example, if your current regression model expresses the outcome in dollars, convert  it to thousands of dollars (divides the values and thus your current regression coefficients by 1000) or even millions of dollars (divides by 1000000). Or choose any factor in between that makes sense.
Expressing results in terms of percentage/fractional changes would best be done by modeling percentage changes directly (e.g., modeling logs of prices, as illustrated in another answer). Whether that makes sense depends on the underlying subject matter. Do you think that an additional bedroom adds a certain number of dollars to the price, or a certain percentage increase to the price? That should determine how you set up your regression.
