Compute the iSAX index for a DataFrame which is assumed to be numeric time series data

h2o.isax(x, num_words, max_cardinality, optimize_card = FALSE)

Arguments

x

an H2OFrame

num_words

Number of iSAX words for the timeseries. ie granularity along the time series

max_cardinality

Maximum cardinality of the iSAX word. Each word can have less than the max

optimize_card

An optimization flag that will find the max cardinality regardless of what is passed in for max_cardinality.

Value

An H2OFrame with the name of time series, string representation of iSAX word, followed by binary representation

References

https://www.cs.ucr.edu/~eamonn/iSAX_2.0.pdf

https://www.cs.ucr.edu/~eamonn/SAX.pdf

Examples

# NOT RUN {
library(h2o)
h2o.init()
df <- h2o.createFrame(rows = 1, cols = 256, randomize = TRUE, value = 0,
                      real_range = 100, categorical_fraction = 0, factors = 0,
                      integer_fraction = 0, integer_range = 100, binary_fraction = 0,
                      binary_ones_fraction = 0, time_fraction = 0, string_fraction = 0,
                      missing_fraction = 0, has_response = FALSE, seed = 123)
df2 <- h2o.cumsum(df, axis = 1)
h2o.isax(df2, num_words = 10, max_cardinality = 10)
# }