穿越/变异很慢，我该如何优化这个调用? [英] crossing / mutate is slow, how could i optimize this call?

问题描述

我有这个电话

distances <- crossing(nodes, gps_points) %>%
    mutate(dist = geosphere::distHaversine(cbind(lon, lat), cbind(rlon, rlat)))

但是需要很长时间才能完成.只是一个交叉函数需要 0.36 秒.整个语句需要 2.587 秒.

But it takes a long time to complete. Just a crossing function takes 0.36 sec. Whole Statement takes 2.587 sec.

是否可以优化此调用?

dput(head(nodes))
structure(list(ids = c(292376151, 1112377287, 6902561109, 5324247975, 
1112377281, 7018492265), ids_igraph = c(128974, 128973, 128972, 
128971, 128970, 128969), lon = c(11.831088, 11.830884, 11.8307377, 
11.8307106, 11.8305447, 11.8303853), lat = c(48.1111416, 48.1108747, 
48.1106855, 48.1106505, 48.1104559, 48.1102703)), row.names = c(833886L, 
2173435L, 4292366L, 4058008L, 2173434L, 4315822L), class = "data.frame")

 dput(head(gps_points))
structure(list(rlon = c(11.7989692, 11.7982155, 11.7968209, 11.7961655, 
11.7957267, 11.7944537), rlat = c(48.0742471, 48.0739067, 48.0733855, 
48.0730923, 48.0728656, 48.0722014)), row.names = c(3250429L, 
3250548L, 323901L, 3250542L, 3250544L, 323902L), class = "data.frame")

您还可以在此处下载 node 和 gps_points 作为 csv 或 dput 文件:

you can also download node and gps_points as csv or dput files here:

节点和 gps_points csv 文件

节点和 gps_points dput 文件

非常感谢，BR.

推荐答案

我将 geosphere::distHaversine 更改为 geodist::geodist_vec() 并设置 measure = "cheap" 以改善运行时间.

I changed geosphere::distHaversine to geodist::geodist_vec() with the setting measure = "cheap" to improve the runtime.

这篇关于穿越/变异很慢，我该如何优化这个调用?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！