There are several ways of specifying arguments to functions in R; in lsfit(urb, infmor), the first argument x corresponds to the independent(s) (matrix or single vector), and the second argument y (dependent variable)
Other more modern modelling functions have a formula interface, for instance with the lm() function, you write: lm(infmor ~ urb + gnpserv) to produce a regression with urb and gnpserv as independents and infmor as dependent variable.
Some examples for formulae you can use. The formula interface lets you write the equations quite naturally, including - as in the third example - the inclusion of interaction terms.
| infmor ~ urb | Bivariate regression infmor dependent |
| infmor ~ urb + gnpserv | Two independents |
| infmor ~ urb + gnpserv+urb*gnpserv | Same with an interaction term |
| infmor ~ urb + continent | continent being a factor variable, R will generate dummy variables for all categories except the first, which will be the reference category |
| infmor~log(urb)) | Of course you can use functions like the log function directly. |
The following symbols are used in a formula:
| Symbol | example | Explanation |
|---|---|---|
| ~ | y~x | Separates dependent from independents |
| + | +x | Add variable X |
| - | -x | Remove variable x |
| * | x*a | Interaction between these variables |
| : | x:a | Include these variables and the interaction between them |
| ^ | (a+b+c)^3 | Include these variables and all interactions up to 3-way |
| poly(x,3) | include a polynomial term | |
| 1 | -1 | Delete the Intercept |
Note that the - sign to remove terms is useful to update a formula (see this document for an example.
See the documentation for further details and more complex model specification.