I tried to use GLM in R, the y value is given below
y <- c(2.0875796, 1.0857121, 0.2783329, 0.7866724, 1.7395036, 1.6341974, 0.1919819, -0.9013408, -1.1337154, 0.4611232, 2.1412645, 2.0390984, 1.8198061, 1.9267886, 1.7026195, 1.6506483, 1.8580555, 1.9983037, 1.9643366, -1.0870642, 0.2191361, 1.9835259, 1.4358765, 2.1582919, 0.5332927, 0.5527285, 1.0580714, 0.3343562, -0.1912542, 1.9449530, 0.5128612, -0.1934113, -0.1820204, -0.4954643, -0.1028118, -0.3829145)
The other two predictor is (let's call it x1 and x2 respectively):
x1 <- c(16, 1, 1, 0, 0, 0, 0, 2, 7, 2, 3, 0, 2, 1, 4, 4, 6, 3, 6, 1, 16, 7, 4, 3, 0, 2, 10, 17, 11, 15, 11, 8, 1, 1, 8, 0)
x2 <- c(2, 19, 18, 19, 18, 19, 20, 17, 12, 16, 15, 18, 16, 17, 14, 14, 12, 15, 12, 20, 0, 11, 14, 15, 19, 16, 8, 1, 7, 3, 7, 10, 18, 17, 10, 19)
When fitting the two variables seperately, the result is kind of expected. However, when I combined those 2 variables (which in this case is highly negatively correlated, with correlation around -0.992), I expected the coefficients of the 2 variables to be more or less similar in terms of magnitude, but the sign is reversed. But in fact, the result gave me high intercept value and the coefficient for both variable is negative which is very weird. Am I missing something here? Thank you in advance
The code I use to build the model is something like below:
dat <- data.frame(x1,x2,y)
model <- glm (y ~ x1 + x2, data = dat)
