inverse cell frequency using standard deviation (SD)

idf_sd(expr, features = NULL, log = FALSE, thres = 0)

Arguments

expr: a matrix, features in row and cells in column
features: vector, feature names or indexes to compute
log: logical, if to do log-transformation
thres: numeric, cell only counts when expr > threshold, default 0

Value

a vector of inverse cell frequency score for each feature

Details

$$\mathbf{IDF_i} = log(1+sd(tf_{i})*\frac{n}{n_i+1})$$ where $tf_i$ is the term frequency of feature $i$, see details in tf(), $n$ is the total number of cells and $n_i$ is the number of cells containing feature $i$.

Examples

data <- matrix(rpois(100, 2), 10, dimnames = list(1:10))
smartid:::idf_sd(data)
#>          1          2          3          4          5          6          7 
#> 0.06191538 0.10035735 0.06957992 0.07781064 0.08512022 0.05712326 0.04502422 
#>          8          9         10 
#> 0.05292170 0.05366416 0.06581973