For each environmental variable, replaces NA, NaN, and
NODATA sentinel values with the mean of the valid values. This
mirrors the Java Maxent partial-data handling where
FeaturedSpace.setSampleExpectations() uses mean substitution
for samples with missing feature values.
Usage
maxent_impute_means(env_vals, nodata_value = -9999)
Arguments
- env_vals
Named list of numeric vectors (one per variable).
- nodata_value
Numeric: sentinel value treated as missing
(default -9999).
Value
Named list of numeric vectors with missing values replaced.
Examples
env <- list(
temp = c(15, NA, 25, 18, -9999),
precip = c(100, 200, 150, 80, 300)
)
filled <- maxent_impute_means(env)
filled$temp[2] # mean of valid values replaces NA
#> [1] 19.33333
