quickfit function

Quick fit of one-step migration

Using k-means clustering to get quick fits of 2 or 3 cluster centers in X-Y coordinates.

quickfit(T, X, Y, dt = 1, n.clust = 2, plotme = TRUE)

Arguments

• T: time
• X: x coordinate of movement
• Y: y coordinate of movement
• dt: duration of migration (arbitrarily = 1)
• n.clust: number of clusters (2 or 3)
• plotme: whether or not to plot the result

Returns

a named vector of initial estimates:

• if n.clust = 2 returns t1, dt, x1, y1, x2, y2
• if n.clust = 3 returns t1, dt1, t2, dt2, x1, y1, x2, y2, x3, y3

Details

This function does estimates the locations and times of migration, but not the duration (dt). It is most useful for obtaining a "null" estimate for seeding the likelihood estimation.

Examples

require(marcher)

## Load simulated data
data(SimulatedTracks)

# plot the MOU simulation
scan_track(MOU.sim)

# quick fit - setting dt = 10
(pm.0 <- with(MOU.sim, quickfit(T, X, Y, dt = 10)))

# interactive locator process
if(interactive()){
 (with(MOU.sim, locate_shift(T, X, Y)))
}

# fit the model
fit <- with(MOU.sim, estimate_shift(T, X, Y))

## Three cluster example

# plot the three range shift simulation
scan_track(MOU.3range)

# quick fit 
## (note - this may not always work!)
with(MOU.3range, quickfit(T, X, Y, dt = 10, n.clust = 3))

if(interactive()){
  with(MOU.3range, locate_shift(T, X, Y, n.clust = 3))
}