Remaining Useful Lifetime Package

Estimate an entities remaining useful lifetime based on survival probabilities

Usage and Configuration

To perform the calculations use the R activity Survival Curves to generate survival probabilities for analyzed entities. A base version of the activity is included in the package. To calculate remaining useful lifetimes, run the workflow Remaining Useful Lifetime. The schedules may be adapted to use different contexts.

Follow the steps given in How do I configure packages pulled from Datagration's package repository to change the configuration of the package.

Details

Generating Survival Curves

To generate survival curves (probability of survival over normalized time), the R activity Survival Curves contained in this package may be used.

This activity performs survival analysis using a Cox Proportional Hazards model. By default, only one predictor signal is used by the script. To add more predictor signals follow these steps:

1. Set up the function to expect additional predictor signals:
   

   survival_analysis <- function(predictorSignal, predictorSignal2, predictorSignal3, survivalTimes) {

2. Make sure the input signals are of type matrix:
   

   predictorSignal <- as.matrix(predictorSignal)
   predictorSignal2 <- as.matrix(predictorSignal2)
   predictorSignal3 <- as.matrix(predictorSignal3)

3. Add the additional inputs to the model:
   

   model <- coxph(Surv(survivalTimes[, entity], events[, entity]) ~ predictorSignal[, entity]+predictorSignal2[, entity]+predictorSignal3[, entity])

After adapting the script, it can be placed in a workflow and run for the entities and scope needed.

The code will return a json string per entity containing the survival curves in json format.

Workflow Remaining Useful Lifetime

This workflow converts survival curves stored in JSON format by the R activity Survival Curves to a reference table (Survival Probabilities) for easier access and uses this table to calculate the remaining useful lifetime based on the survival probability curve and a user-specified probability threshold.

Reviewing Survival Curves

To ease the reviewing process, the package contains the P# script Survival Curve Visualization as well as the chart style Survival Curve. Using those the survival curves can either be shown in the Develop page or on a dashboard as shown below.

Data Requirements

The following data is required:

runtime between failures (as calculated by the Runtime Between Failures package)
predictor signals (time dependent signals, which are indicative of a pending failure)