skewness and kurtosis aren’t called on the data (array) directly. You have to fit! a statistical model at first:
https://joshday.github.io/OnlineStats.jl/stable/stats_and_models/#Univariate-Statistics-1
julia> data=randn(1000);
julia> mean(data)
-0.029101214826974715
julia> var(data)
0.9627532944591485
julia> o = fit!(Moments(),data)
Moments: n=1000 | value=[-0.0291012, 0.962637, -0.073216, 2.84648]
julia> mean(o)
-0.029101214826974704
julia> var(o)
0.9627532944591476
julia> std(o)
0.9811999258352742
julia> skewness(o)
0.011425034687808195
julia> kurtosis(o)
0.07320689858578566