Survival Analysis

Telco Data: Kaplan-Meier (KM) estimators

StatsResource

Survival Analysis

Here we will apply Kaplan-Meier (KM) estimators and smoothed hazard functions to the telco data set.

Key Exercises

  1. Kaplan-Meier Estimator for Telco Churn Data:
    • Construct the KM estimator for the telco churn data.
    • Find the median and mean survival times for this data.
    • Repeat these steps for the prostate survival and smoker data.
  2. Follow-Up Time Measurement:
    • Measure the follow-up time by reversing the censoring labels in the dataset.
  3. Smoothed Hazard Functions:
    • Construct smoothed hazard estimators for the telco churn data.
    • Determine the times with the highest and lowest hazard rates.
    • Repeat these steps for the prostate survival and smoker data.
  4. Boundary Corrections:
    • Explore boundary corrections in smoothed hazard functions using the muhaz() function.
  5. Comparing KM and Smoothed Estimates:
    • Compare the Kaplan-Meier estimates and smoothed hazard estimates by integrating the hazard function to calculate the survival function.
    • Evaluate the differences with and without boundary corrections.
  6. Estimating Survival Curves for Telco Churn Data:
    • Build KM estimates for the telco churn data and splits based on vmailplan and intlplan.
    • Plot the survival curves and analyze differences using confidence bounds.
  7. Combining KM Estimators:
    • Stack KM estimator data for combined visualization.
    • Simplify and compare curves for different subgroups to identify patterns and insights.

Next Steps

Participants can further explore the practical application of these survival analysis techniques to gain deeper insights into their datasets. These exercises provide a comprehensive understanding of survival analysis, from basic concepts to advanced estimations and visualizations.