R/RDring_coef.R

In mengluchu/APMtools: statistical air pollution mapping tools

#' Plot coefficients of roadrings of fitting a linear regression model.
#' @param variabledf dataframe including predictors and dependent variable
#' @param y_var the quote of dependent variable name (e.g. quote (value_mean))
#' @param Road_varname the name of variables contains road buffers, e.g. 'ROAD'
#' @param pop_var other variables added to linear regression, in a list or vector if multiple. e.g. popvar = c('pop3k', 'RSp', 'elevation')
#' @param number_roadtypes number of roadtypes, by default 3
#' @param buffers_in buffers of the inner ring
#' @param buffers_out buffers of the outer ring
#' @export

RDring_coef = function(variabledf, y_var, Road_varname = "ROAD", number_roadtypes = 3, buffers_in, buffers_out, pop_var = NA, normalize = F) {
    
    # substitute(y_var)
    
    
    rdf2 = create_ring(variabledf, normalize = normalize, number_roadtypes = number_roadtypes, buffers_in = buffers_in, buffers_out = buffers_out)
    
    if (!is.na(pop_var)) {
        
        addiv = paste("|", paste(pop_var, collapse = "|"), sep = "")
        grepstring = paste(Road_varname, "|", deparse(y_var), addiv, sep = "")
    } else grepstring = paste(Road_varname, "|", deparse(y_var), sep = "")
    
    pre_mat = subset_grep(rdf2, grepstring)
    f = formula(paste(y_var, "~.", sep = ""))
    alm = lm(f, data = pre_mat)
    # print( summary(alm))
    allcoef = coef(alm)[which(grepl(Road_varname, names(coef(alm))))]  # take off 1,(2 if there is another parameter)
    
    n_buffer = length(buffers_in)
    shortname = gsub("^([^_]*_[^_]*)_.*$", "\\1", names(allcoef))
    coefdf = data.frame(bufferring = rep(c(1:n_buffer), length(allcoef)/n_buffer), coeffi = allcoef, fs = shortname)
    ggplot(data = coefdf, aes(x = bufferring, y = coeffi)) + geom_line() + facet_grid(facets = fs ~ ., scales = "free")
}