Data Science on the Google Cloud Platform by Unknown

Author:Unknown
Language: eng
Format: epub
Published: 0101-01-01T00:00:00+00:00

To choose the variables in the dataset that we can use in our prediction problem, we have to consider whether the variable is related to the problem at hand and whether it will violate causality constraints or incur a training-serving skew. Based on these considerations, we selected 27 fields to download.

There are three possible data processing architectures on the cloud for large datasets -- scaling up, scaling out, and data in situ. Scaling up is very efficient, but is limited by the size of the largest machine you can get a hold of. Scaling out is very popular but requires that you pre-shard your data by splitting it among compute nodes and thus leads to maintaining expensive clusters unless you can hit sustained high utilization. Keeping data in situ is possible only if your datacenter supports petabits per second of bisectional bandwidth so that any file can be moved to any compute node in the datacenter on demand. Because Google Cloud has this capability, we will upload our data to Google Cloud Storage (GCS), a blob storage that is not pre-sharded.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.