map inside map2 - how to refer properly to arguments (purrr)

Question

ex <- structure(list(group = c("group B", "group B", "group C", "group B","group C", "group B", "group B", "group A", "group C", "group C", "group C", "group B", "group A", "group A", "group A", "group B", "group A", "group A", "group B", "group C", "group B", "group A", "group C", "group C", "group C", "group C", "group B", "group A", "group A", "group C", "group B", "group A", "group A", "group B", "group C", "group C", "group A", "group C", "group C", "group A", "group B", "group B", "group A", "group B", "group C", "group C","group A", "group B", "group C", "group C"), A1 = c(0.765913072274998, 0.167720616329461, 0.282011203467846, 0.16467465297319, 0.407501850277185, 0.33958561392501, 0.117573569528759, 0.267871993361041, 0.930967768887058, 0.286146199563518, 0.741841563722119, 0.637853658990934, 0.137378493556753, 0.820813736645505, 0.249520575627685, 0.275153698632494, 0.916794545250013, 0.316050065914169, 0.393918378278613, 0.342175736324862, 0.0177193265408278, 0.178873546421528, 0.376545072998852, 0.411527326330543, 0.904074088903144, 0.487975180381909, 0.491365089081228, 0.591370195383206, 0.319207336986437, 0.98943907325156, 0.916014631278813, 0.0347612821497023, 0.323899461887777, 0.155270972754806, 0.436683354899287, 0.316902073565871, 0.734995431266725, 0.584133808733895, 0.515310257440433, 0.921727291075513, 0.0689518100116402, 0.659549278207123, 0.894137248862535, 0.00174906081520021, 0.873320956015959, 0.77207364118658, 0.637504813494161, 0.473099726485088, 0.557896945858374, 0.632965805241838), A2 = c(0.782154354499653, 0.718993512215093, 0.391234505455941, 0.337346265325323, 0.141482090810314, 0.587817938998342, 0.384924706770107, 0.0679492244962603, 0.0509498412720859, 0.786300176288933, 0.00685039279051125, 0.361857839627191, 0.851737944642082, 0.333896369440481, 0.521961389342323, 0.761324436869472, 0.486214824952185, 0.249763275263831, 0.536617708392441, 0.982582966331393, 0.879302836721763, 0.0212801641318947, 0.999207010492682, 0.661623647902161, 0.514440550701693, 0.748157452791929, 0.609151393873617, 0.581557413795963, 0.495366840157658, 0.595225095050409, 0.694380027009174, 0.419036868494004, 0.618371620541438, 0.406731882831082, 0.947823651600629, 0.182527825701982, 0.365398081485182, 0.307149735512212, 0.905119536910206, 0.657605888554826, 0.706386201782152, 0.461993521312252, 0.637554163113236, 0.280387100065127, 0.454221101710573, 0.0712104975245893, 0.914795317919925, 0.951028517214581, 0.645093881059438, 0.754043457563967), A3 = c(0.590488174697384, 0.876135899219662, 0.349565496202558, 0.365676332963631, 0.709230658365414, 0.584304825868458, 0.391973132034764, 0.464247716590762, 0.00831679091788828, 0.282901889178902, 0.842566592851654, 0.141866789199412, 0.278708242345601, 0.680587171344087, 0.256092368392274, 0.535304376389831, 0.803430012892932, 0.336343225324526, 0.320332229137421, 0.809689761372283, 0.588527292944491, 0.767302295425907, 0.124350237427279, 0.605355758452788, 0.619420127244666, 0.326774680987, 0.917224677512422, 0.710018905811012, 0.892817938234657, 0.149181636283174, 0.65066168922931, 0.433064805110916, 0.167979725869372, 0.809581968234852, 0.803237372776493, 0.703188817715272, 0.507392750121653, 0.372131450567394, 0.0688441153615713, 0.928956841118634, 0.960712827509269, 0.37454927386716, 0.753415656508878, 0.687665716046467, 0.05052674934268, 0.155349446227774, 0.806162646971643, 0.725155076943338, 0.537310504587367, 0.674253351520747), A4 = c(0.426875792676583, 0.168233293108642, 0.38692078506574, 0.673673333134502, 0.221049380488694, 0.142470651771873, 0.505352358799428, 0.579006788786501, 0.809476702939719, 0.343090934911743, 0.136329119792208, 0.881694708252326, 0.142607795307413, 0.658202062360942, 0.0624804550316185, 0.938871977152303, 0.477995269699022, 0.989794839406386, 0.307003591908142, 0.40553830191493, 0.0249065780080855, 0.321581491269171, 0.432656849268824, 0.578710418893024, 0.482647196389735, 0.72430428257212, 0.611029474530369, 0.748521578731015, 0.939656358910725, 0.803305297158659, 0.339922665851191, 0.919090943178162, 0.0926963407546282, 0.671128012472764, 0.634122629882768, 0.219061656622216, 0.376445228001103, 0.468331813113764, 0.131768246181309, 0.258267979836091, 0.651934198103845, 0.678243630565703, 0.663701833924279, 0.678762876661494, 0.524524878012016, 0.380242201732472, 0.433922954136506, 0.795754680642858, 0.383180371485651, 0.160383063135669)), .Names = c("group", "A1", "A2", "A3", "A4"), row.names = c(NA, -50L), class = c("tbl_df", "tbl", "data.frame"))

With above sample data I want to perform msClustering within groups. This clustering requires tuning parameter h so I define few values of it in column h.cand. Then I want to call msClustering with subsequent values of h and store the output in a list column. Theoretically, it should be feasible with purrr, but I think it requires nested map, and precisely speaking map inside map2. Here is my problem, I'm not sure how to refer for different list arguments. I have tried something like below:

ex %>% 
  group_by(group) %>% 
  nest() %>% 
  h.cand = map(data, ~quantile(dist(.x), seq(0.05, 0.40, by = 0.05))) %>% 
  mutate(cluster = map2(h.cand, data, ~map(.x, ~msClustering(
    .y, # data (second argument of outter map2)
    h = .x # h.cand element (first argument of inner map)
  ))))

and ended up with error:

Error: cannot allocate vector of size 1681.9 Gb

How should I refer to elements of outter and inner map in order to perform 8 (a length of h.cand vector) clusterings for each group?

Answer 1

For complicated anonymous functions, like this one, it's better if you use the function(x) instead of lambda/~ syntax for passing to map()'s .f argument.

Clean up the data:

map(ex, length)
# make element5 same length
ex[[5]] <- c(ex[[5]], runif(16))
# make into data frame
ex <- dplyr::bind_cols(ex) 

Use function(x) instead of ~:

ex2 <- ex %>% 
    group_by(group) %>% 
    nest() %>%
    mutate(h.cand = map(data,
                        ~ quantile(dist(.), seq(0.05, 0.40, by = 0.05))),
           cluster = map2(h.cand, data,
                          function(x, y) { map(x,
                                               function(x2) { msClustering(y, x2) }) } ) )

Result check:

unnest(ex2, cluster)
# A tibble: 24 x 2
   group   cluster   
   <chr>   <list>    
 1 group B <list [2]>
 2 group B <list [2]>
 3 group B <list [2]>
 4 group B <list [2]>
 5 group B <list [2]>
 6 group B <list [2]>
 7 group B <list [2]>
 8 group B <list [2]>
 9 group C <list [2]>
10 group C <list [2]>
# ... with 14 more rows