我做了一个简单的例子，我想做的事情，我打算更灵活。
我希望能够在h2o.frame的行上进行子集，对这些行进行一些计算，然后将结果分配给这些相同的行。在这个例子中，我计算了相对的"mpg"在每组" cycle "
library(h2o)
packageVersion("h2o")
[1] ‘3.32.1.3’
version
_
platform x86_64-w64-mingw32
arch x86_64
os mingw32
system x86_64, mingw32
status
major 4
minor 1.1
year 2021
month 08
day 10
svn rev 80725
language R
version.string R version 4.1.1 (2021-08-10)
nickname Kick Things
h2o.init()
mtcars <- as.h2o(mtcars)
mtcars$mpg_rel <- NA
mtcars
mpg cyl disp hp drat wt qsec vs am gear carb mpg_rel
1 21.0 6 160 110 3.90 2.620 16.46 0 1 4 4 NaN
2 21.0 6 160 110 3.90 2.875 17.02 0 1 4 4 NaN
3 22.8 4 108 93 3.85 2.320 18.61 1 1 4 1 NaN
4 21.4 6 258 110 3.08 3.215 19.44 1 0 3 1 NaN
5 18.7 8 360 175 3.15 3.440 17.02 0 0 3 2 NaN
6 18.1 6 225 105 2.76 3.460 20.22 1 0 3 1 NaN
[32 rows x 12 columns]
mtcars[mtcars[["cyl"]] == 4, "mpg"] / h2o.mean(mtcars[mtcars[["cyl"]] == 4, "mpg"])
mpg
1 0.8550972
2 0.9151040
3 0.8550972
4 1.2151381
5 1.1401296
6 1.2713945
[11 rows x 1 column]
# however, the assignment throws an error and corrupts `mtcars`
mtcars[mtcars[["cyl"]] == 4, "mpg_rel"] <- mtcars[mtcars[["cyl"]] == 4, "mpg"] / h2o.mean(mtcars[mtcars[["cyl"]] == 4, "mpg"])
ERROR: Unexpected HTTP Status code: 412 Precondition Failed (url = http://localhost:54321/99/Rapids)
water.exceptions.H2OIllegalArgumentException
[1] "water.exceptions.H2OIllegalArgumentException: unimplemented"
[2] " water.H2O.unimpl(H2O.java:1310)"
[3] " water.rapids.ast.prims.assign.AstRectangleAssign.apply(AstRectangleAssign.java:93)"
[4] " water.rapids.ast.prims.assign.AstRectangleAssign.apply(AstRectangleAssign.java:30)"
[5] " water.rapids.ast.AstExec.exec(AstExec.java:63)"
[6] " water.rapids.ast.prims.assign.AstTmpAssign.apply(AstTmpAssign.java:48)"
[7] " water.rapids.ast.prims.assign.AstTmpAssign.apply(AstTmpAssign.java:17)"
[8] " water.rapids.ast.AstExec.exec(AstExec.java:63)"
[9] " water.rapids.Session.exec(Session.java:85)"
[10] " water.rapids.Rapids.exec(Rapids.java:94)"
[11] " water.api.RapidsHandler.exec(RapidsHandler.java:38)"
[12] " java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)"
[13] " java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)"
[14] " java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)"
[15] " java.base/java.lang.reflect.Method.invoke(Method.java:566)"
[16] " water.api.Handler.handle(Handler.java:60)"
[17] " water.api.RequestServer.serve(RequestServer.java:470)"
[18] " water.api.RequestServer.doGeneric(RequestServer.java:301)"
[19] " water.api.RequestServer.doPost(RequestServer.java:227)"
[20] " javax.servlet.http.HttpServlet.service(HttpServlet.java:707)"
[21] " javax.servlet.http.HttpServlet.service(HttpServlet.java:790)"
[22] " org.eclipse.jetty.servlet.ServletHolder.handle(ServletHolder.java:865)"
[23] " org.eclipse.jetty.servlet.ServletHandler.doHandle(ServletHandler.java:535)"
[24] " org.eclipse.jetty.server.handler.ScopedHandler.nextHandle(ScopedHandler.java:255)"
[25] " org.eclipse.jetty.server.handler.ContextHandler.doHandle(ContextHandler.java:1317)"
[26] " org.eclipse.jetty.server.handler.ScopedHandler.nextScope(ScopedHandler.java:203)"
[27] " org.eclipse.jetty.servlet.ServletHandler.doScope(ServletHandler.java:473)"
[28] " org.eclipse.jetty.server.handler.ScopedHandler.nextScope(ScopedHandler.java:201)"
[29] " org.eclipse.jetty.server.handler.ContextHandler.doScope(ContextHandler.java:1219)"
[30] " org.eclipse.jetty.server.handler.ScopedHandler.handle(ScopedHandler.java:144)"
[31] " org.eclipse.jetty.server.handler.HandlerCollection.handle(HandlerCollection.java:126)"
[32] " org.eclipse.jetty.server.handler.HandlerWrapper.handle(HandlerWrapper.java:132)"
[33] " water.webserver.jetty9.Jetty9ServerAdapter$LoginHandler.handle(Jetty9ServerAdapter.java:130)"
[34] " org.eclipse.jetty.server.handler.HandlerCollection.handle(HandlerCollection.java:126)"
[35] " org.eclipse.jetty.server.handler.HandlerWrapper.handle(HandlerWrapper.java:132)"
[36] " org.eclipse.jetty.server.Server.handle(Server.java:531)"
[37] " org.eclipse.jetty.server.HttpChannel.handle(HttpChannel.java:352)"
[38] " org.eclipse.jetty.server.HttpConnection.onFillable(HttpConnection.java:260)"
[39] " org.eclipse.jetty.io.AbstractConnection$ReadCallback.succeeded(AbstractConnection.java:281)"
[40] " org.eclipse.jetty.io.FillInterest.fillable(FillInterest.java:102)"
[41] " org.eclipse.jetty.io.ChannelEndPoint$2.run(ChannelEndPoint.java:118)"
[42] " org.eclipse.jetty.util.thread.strategy.EatWhatYouKill.runTask(EatWhatYouKill.java:333)"
[43] " org.eclipse.jetty.util.thread.strategy.EatWhatYouKill.doProduce(EatWhatYouKill.java:310)"
[44] " org.eclipse.jetty.util.thread.strategy.EatWhatYouKill.tryProduce(EatWhatYouKill.java:168)"
[45] " org.eclipse.jetty.util.thread.strategy.EatWhatYouKill.run(EatWhatYouKill.java:126)"
[46] " org.eclipse.jetty.util.thread.ReservedThreadExecutor$ReservedThread.run(ReservedThreadExecutor.java:366)"
[47] " org.eclipse.jetty.util.thread.QueuedThreadPool.runJob(QueuedThreadPool.java:762)"
[48] " org.eclipse.jetty.util.thread.QueuedThreadPool$2.run(QueuedThreadPool.java:680)"
[49] " java.base/java.lang.Thread.run(Thread.java:834)"
Error in .h2o.doSafeREST(h2oRestApiVersion = h2oRestApiVersion, urlSuffix = page, :
ERROR MESSAGE:
unimplemented
我发现的一个解决方案是使用行号的vector来进行子集。但是，当使用大数据时，as.vector转换效率非常低。如果能有类似上述方法的方法，那就太好了。
which_rows <- as.vector(h2o.which(mtcars[["cyl"]] == 4))
mtcars[which_rows, "mpg_rel"] <- mtcars[which_rows, "mpg"] / h2o.mean(mtcars[which_rows, "mpg"])
mtcars
mpg cyl disp hp drat wt qsec vs am gear carb mpg_rel
1 21.0 6 160 110 3.90 2.620 16.46 0 1 4 4 NaN
2 21.0 6 160 110 3.90 2.875 17.02 0 1 4 4 NaN
3 22.8 4 108 93 3.85 2.320 18.61 1 1 4 1 0.8550972
4 21.4 6 258 110 3.08 3.215 19.44 1 0 3 1 NaN
5 18.7 8 360 175 3.15 3.440 17.02 0 0 3 2 NaN
6 18.1 6 225 105 2.76 3.460 20.22 1 0 3 1 NaN
[32 rows x 12 columns]
我不确定这是否是最好的方法，但使用以下方法应该比使用as.vector更有效(一次传递所有行以获得mpg_mean和一次传递h2o.ifelse)。
mpg_mean <- h2o.mean(mtcars[mtcars[["cyl"]] == 4, "mpg"])
mtcars[["mpg_rel"]] <- h2o.ifelse(mtcars[["cyl"]] == 4, mtcars[["mpg"]] / mpg_mean, mtcars[["mpg"]])