Toggle navigation
Home
▼ Details
Products and pricing
Chart gallery
User stories
Text analytics
CDC NAMCS Library
Blog
Tutorials
Contact
Sign in
Post Editor
← All blog posts
View post
Save
<p>Surveys often ask for time intervals, with start and end dates:</p> <ul> <li>When did you buy the product? When did you finish it?</li> <li>When did the patient start and end each line of therapy?</li> <li>When did respondents start and end different programs?</li> </ul> <p>One thing we can do is to look at the data. Another is to look at how survival data is summarized in clinical research...</p> <p><img src="/images/blog/2018-07-18-kaplan/intervals-toyota.png" style="max-width: 540px; margin-top:20px"/></p> <!--more--> <h2 id="visualizing-intervals">Visualizing intervals</h2> <p>The chart above is visualization of production runs for a number of Toyota models using Protobi's <code>IntervalChart</code>. </p> <p>Each line represents a single model, with start and stop dates:</p> <ul> <li>The Toyota Land Cruiser was first produced in 1951, and is still in production today, 67 years later.</li> <li>The Toyota Stout was first produced in 1954, but stopped in 1989 after 35 years.</li> </ul> <p>Some runs are complete, these we show these with round endpoints. Some runs are still continuing at the time of this analysis, these are shown cut off with a straight end.</p> <p>Here the runs are sorted in ascending order of start date. These could be ordered by end date, duration run, etc.</p> <p>Scanning a chart like this is useful to just see what the data looks like.</p> <h2 id="distribution-of-durations-to-date">Distribution of durations to-date</h2> <p>We can calculate the duration for each automobile model, and display it as a statistical distribution. </p> <p>Here we can differentiate between production runs that are complete (for whom the data is final) versus those intervals that are still ongoing.</p> <p><img src="/images/blog/2018-07-18-kaplan/production-duration.png" style="max-width: 540px"/></p> <p>Here we can clearly see that many production runs are still ongoing. So a simple mean or median of run lengths to-date would seriously under-report the expected production runs.</p> <p>For right censored data, a useful way to summarize the data is the <a href="/post/kaplan-meier-survival-estimator">Kaplan-Meier Survival Estimator</a></p>
Date
Status
Published
Draft
Slug
edit
Thumbnail
Categories
Manage
Release
Features
Datasets
Surveys
Tips
NAMCS
Applications
Crosstab
Tutorial
Design
Concepts
Segmentation
Examples
Blog Test Category
Delete
Convert to MD