Linear regression model of log variable conc over time

[Goh jedrek]

I have to perform a linear regression of conc (natural log) over time separatedly for each participant (id) and diet. (eg participant 1, diet = meat; participant 1, diet = nomeat; participant 2, diet = meat, etc.)

I am given the dataset for the coefficient for variable time from linear regression output (times -1), known as ext and predicted concentration from linear regression output, known as pre.

I am supposed to generate ext_2 and pre_2 to check that ext and pre are calculated correctly.

The equation of the linear regression is as follows: ln conc = beta naught + beta one x time

beta one = ext variable

So far I have tried the following:

gen lnconc = ln(conc)

> Generate the natural log variable for conc

regress lnalcohol time id if diet == "meat"
            regress lnalcohol time id if diet == "nomeat"

> Fit linear regression model (where I am stucked at)

predict pre_2, xb

> predicted conc for the output, but does not return the same values as shown in pre

matrix a = r(table)
local est = a[a,b];

> ext variable for coefficient, but had a syntax error when i tried the second line of code (also stucked at this)

Looking forward to some advice~