# Datasets¶

Datasets

This file contains several datasets that are useful for testing and experimenting To import you can use the following format:

from reliability.Datasets import automotive failures = automotive().failures right_censored = automotive().right_censored
from reliability.Datasets import automotive print(automotive().info) print(help(automotive))
class reliability.Datasets.ALT_load

This is an accelerated life test (ALT) dataset conducted at 3 loads It should be used with an ALT probability plot This is a relatively small dataset with just 20 and no censoring Sourced from Dr. Mohammad Modarres, University of Maryland

class reliability.Datasets.ALT_load2

This is an accelerated life test (ALT) dataset conducted at 3 loads It should be used with an ALT probability plot This is a relatively small dataset with just 18 values, 5 of which are censored. Sourced from Dr. Mohammad Modarres, University of Maryland

class reliability.Datasets.ALT_temperature

This is an accelerated life test (ALT) dataset conducted at 3 temperatures It should be used with an ALT probability plot The dataset contains mostly censored data but is easily fitted by Weibull_2P, Lognormal_2P, and Gamma_2P distributions. Normal_2P will fit but the ALT probability plot will not show Normal_2P is a good fit for this dataset Sourced from Dr. Mohammad Modarres, University of Maryland

class reliability.Datasets.ALT_temperature2

This is an accelerated life test (ALT) dataset conducted at 4 temperatures It should be used with an ALT probability plot This is a relatively small dataset with just 40 values, 20 of which are censored. Sourced from Dr. Mohammad Modarres, University of Maryland

class reliability.Datasets.ALT_temperature3

This is an accelerated life test (ALT) dataset conducted at 3 temperatures It should be used with an ALT probability plot This is a relatively small dataset with just 30 values, none of which are censored.

class reliability.Datasets.ALT_temperature4

This is an accelerated life test (ALT) dataset conducted at 3 temperatures It should be used with an ALT probability plot This is a relatively small dataset with just 20 values, none of which are censored.

class reliability.Datasets.ALT_temperature_humidity

This is an accelerated life test (ALT) dataset conducted at 2 different temperatures and 2 different humidities The dataset is fairly small but has no censored data It is recommended to use a dual-stress model such as Dual-Exponential model

class reliability.Datasets.ALT_temperature_voltage

This is an accelerated life test (ALT) dataset conducted at 2 different temperatures and 2 different voltages The dataset is small but contains no censored values. It is recommended to use a dual-stress model such as the Power-Exponential model.

class reliability.Datasets.ALT_temperature_voltage2

This is an accelerated life test (ALT) dataset conducted at 3 different temperatures and 2 different voltages The dataset is fairly small and the pattern of stresses make it extremely difficult to fit. Note that there is stress-pair that contains only a single failure. It is recommended to use a dual-stress model.

class reliability.Datasets.MCF_1

This dataset is formatted for use with the Mean Cumulative Function (MCF_parametric or MCF_nonparametric) It consists of failure times for 5 systems. It exhibits a fairly constant failure rate, appearing to be slightly increasing (beta > 1)

class reliability.Datasets.MCF_2

This dataset is formatted for use with the Mean Cumulative Function (MCF_parametric or MCF_nonparametric) It consists of failure times for 56 systems. It exhibits an increasing failure rate at the start and a decreasing failure rate near the end. Due to this shape it is not fitted well by the power law model used in MCF parametric.

class reliability.Datasets.automotive

This dataset is relatively small and a challenging task to fit with any distribution due to its size and shape It also includes mostly right censored data which makes fitting more difficult. Sourced (with permission) from: V.V. Krivtsov and J. W. Case (1999), Peculiarities of Censored Data Analysis in Automotive Industry Applications - SAE Technical Paper Series, # 1999-01-3220

class reliability.Datasets.defective_sample

This dataset is heavily right censored with intermixed censoring (not all censored values are greater than the largest failure) It exhibits the behavior of a defective sample (aka. Limited fraction defective) Thanks to Alexander Davis for providing this dataset.

class reliability.Datasets.electronics

This dataset is heavily right censored without intermixed censoring (all censored values are greater than the largest failure) It is very difficult to fit and requires a specific combination of initial guess (least squares) and optimizer (TNC) to achieve the lowest log-likelihood. Thanks to Jiwon Cha for providing this dataset.

class reliability.Datasets.mileage

This dataset is simple to fit. It contains 100 values with no right censoring. The data appears to be from a Normal Distribution. Sourced from Example 2.31 (page 63) of Reliability Engineering and Risk analysis 3rd Edition by M. Modarres, M. Kaminskiy, and V.V. Krivtsov

class reliability.Datasets.mixture

This dataset is a mixture model with heavy censoring (97.90622%) It is best modelled using a Weibull Mixture Model.