dynamichazard (version 0.5.1)

hds: Hard Drive failures.

Description

A data set containing hard drive failures data from Backblaze in the start-stop format used in the survival package.

Usage

hds

Arguments

Format

A data.frame with the following columns:

serial_number

Serial number for the hard disk which the row belongs to.

model

hard disk model.

manufacturer

manufacturer of the hard disk model.

tstart,tstop

start and stop times on the SMART 9 attribute scale.

fails

1 if the hard disk fails at tstop.

size_tb

hard disk size in terabytes.

smart_x

the raw SMART attribute x value. E.g., smart_12 is the power cycle count.

smart_x_bin

1 if the SMART attribute x value is non-zero.

..._cumsum

cumulative sum of the prefix ....

n_fails

number of failures in the original data. Hard disk should only fail once but this is not the case in the raw data.

n_records

number of records in the original source.

min_date,max_date

first and last date in the original source.

min_hours,max_hours

smallest and largest value of the SMART 9 attribute in the original source.

Details

Details about the the SMART attributes can be found on https://en.wikipedia.org/wiki/S.M.A.R.T.. As stated in the original source

"Reported stats for the same SMART stat can vary in meaning based on the drive manufacturer and the drive model. Make sure you are comparing apples-to-apples as drive manufacturers don't generally disclose what their specific numbers mean."

There are some notes on https://en.wikipedia.org/wiki/S.M.A.R.T. regarding which attributes that have vendor specific raw value. Further,

"The values in the files are the values reported by the drives. Sometimes, those values are out of whack. For example, in a few cases the RAW value of SMART 9 (Drive life in hours) reported a value that would make a drive 10+ years old, which was not possible. In other words, it<U+2019>s a good idea to have bounds checks when you process the data."

Last value carried forward have been used for all covariates. See this github page for the processing steps https://github.com/boennecd/backblaze_survival_analysis_prep.