mannkendall {emon}

Mann-kendall test for trend

Description

Calculates the Mann-Kendall statistic for monotonic trend and also the p-value against the null hypothesis of no trend. Unlike the function MannKendall, works for repeat values of time.

Usage

 mannkendall(time, Y, nsims.mk=999) 

Arguments

time

Vector of values which define the direction of the trend.

Y

Vector of values for which you want to determine the trend.

nsims.mk

Number of replicate permuations to calculate the p-value. Default=999.

Details

Error checks of parameters not included so as not to slow down mannkendall. The statistic is calculated by considering each case j and considering the subset of observations that have time greater than time[j]. The Mann Kendall statistic is the number of observations in this subset for which Y > Y[j] minus the number for which Y < Y[j]. The statistic is summed over all j. The p-value is calculated by nreps random permutations of the Y values.

Value

mann

Mann-Kendall statistic of observed data, as calculated by mannkendall.stat.

pvalue

P-value assuming a null hypothesis of no trend and two-way alternative hypothesis.

Author(s)

Jon Barry: [e-mail]

References

Mann, H.B. (1945), Nonparametric tests against trend, Econometrica, 13, 245-259.

Kendall, M.G. 1975. Rank Correlation Methods, 4th edition, Charles Griffin, London.

See Also

mannkendall.stat, power.trend, MannKendall

Examples

x = rep(1:10,rep(2,10))
y = rnorm(20, 5, 2)
mannkendall(x, y)