For each ordered pair of channels, how many converting journeys contain the first channel somewhere before the second. Read a row as "this channel assisted these channels".
Arguments
- paths
An
mm_pathsobject frombuild_paths().- normalise
Divide each cell by the number of converting journeys containing the assisting channel, giving a rate rather than a count? Defaults to
FALSE.
Details
The diagonal counts journeys where a channel appears before itself, which
is repeat exposure. Under build_paths()'s default
collapse_repeats = TRUE, consecutive repeats have already been merged, so
the diagonal only picks up channels that recur after an intervening
different channel. Build the paths with collapse_repeats = FALSE if you
want back-to-back repeat exposure to show up here.
Examples
data(mm_events)
paths <- build_paths(mm_events, id = "customer_id", channel = "channel",
timestamp = "timestamp", conversion = "conversion")
round(assisted_conversions(paths, normalise = TRUE), 3)
#> assisted
#> assisting (blank) (missing) (none) affiliate direct display email
#> (blank) 0.000 0.009 0.000 0.078 0.000 0.114 0.192
#> (missing) 0.036 0.000 0.036 0.116 0.009 0.125 0.143
#> (none) 0.000 0.005 0.000 0.137 0.000 0.141 0.185
#> affiliate 0.041 0.015 0.043 0.091 0.034 0.160 0.252
#> direct 0.000 0.006 0.000 0.094 0.000 0.106 0.159
#> display 0.045 0.023 0.047 0.163 0.038 0.101 0.303
#> email 0.021 0.013 0.017 0.078 0.021 0.090 0.068
#> organic_search 0.031 0.019 0.028 0.119 0.020 0.135 0.225
#> paid_search 0.027 0.013 0.022 0.089 0.025 0.096 0.179
#> social 0.051 0.023 0.041 0.136 0.036 0.176 0.288
#> assisted
#> assisting organic_search paid_search social
#> (blank) 0.205 0.237 0.123
#> (missing) 0.143 0.259 0.134
#> (none) 0.171 0.263 0.137
#> affiliate 0.281 0.342 0.191
#> direct 0.212 0.259 0.153
#> display 0.316 0.415 0.259
#> email 0.166 0.250 0.128
#> organic_search 0.125 0.296 0.167
#> paid_search 0.191 0.128 0.125
#> social 0.267 0.391 0.117