
上QQ阅读APP看书,第一时间看更新
Getting ready
The previous chapter provided the details for the installation of H2O in R along with a working example using its web interface. To start modeling, load the h20 package in the R environment:
require(h2o)
Then, initialize a single-node H2O instance using the h2o.init() function on eight cores and instantiate the corresponding client module on the IP address localhost and port number 54321:
localH2O = h2o.init(ip = "localhost", port = 54321, startH2O = TRUE,min_mem_size = "20G",nthreads = 8)
The H2O package has dependency on the Java JRE. Thus, it should be pre-installed before executing the initialization command.