drop_labels

Advances in Financial Machine Learning, Snippet 3.8 page 54. This function recursively eliminates rare observations.

drop_labels(events, min_pct = 0.05)

Arguments

events	Events with labels
min_pct	A fraction used to decide if the observation occurs less than that fraction.

Examples

data("spy")
close <- subset(spy, select = c("index", "close"))
daily_vol <- daily_volatility(close)
cusum_events <- cusum_filter(close, 0.001)
vertical_barriers <- add_vertical_barrier(cusum_events, spy$index, num_days = 1)
events <- get_events(price = close,
                     t_events = cusum_events,
                     pt_sl = c(1, 2),
                     target = daily_vol,
                     min_ret = 0.005,
                     vertical_barrier_times=vertical_barriers,
                     side_prediction=NA)
labels <- get_bins(events, close)
labels_red <- drop_labels(labels, 0.05)