Time Series

See all 4+ dimensions (time) in visualizations

Overview

Excellence defined
Fortune 500 in the US
Hans Roling
Cycle Time
H20.ai
Time Series databases
More on front-end styling

Understanding the trend of numbers over time is a fundamental skill. That’s why it’s taught in elementary school. Why?

“Those who don’t know history are doomed to repeat it.” – Edmund Burke

Excellence defined

Below are vizualizations that have the following traits:

Dynamic data - update visualizations (in “live mode”) as data changes in sources such as databases.
Visual querying - change the query by selecting or clicking on a portion of the graph or chart (to drill down, for example).
Linked multi-dimensional visualization - selections made in one chart are reflected as you navigate into other charts.
Animation -

PROTIP: Dynamic (movie mode) is available only within the Tableau Public client, not when viewed on websites (as of 2016-01-05).

Personalization - give power users an in-depth view and newbies a simpler view, and also control access to data based on user- and role-based access privileges.
Actionable alerts - thresholds and parameters that trigger messages whether you’re interacting with reports or not.

Fortune 500 in the US

View https://fortune.com/fortune500/visualizations/
to select an industry and mouse over a line to see each company’s ranking change over the last 20 years.

The higher each line appears, the higher the company is on Fortune magazine’s 500 largest public companies.

Companies that have a rising trajectory include:

Whole Foods Market
Amazon
Alphabet (Google)
Apple
CVS Health
Comcast
Dollar General
Gilead Sciences
Qualcomm
Facebook
Visa
Master Card
Cognizant
Oracle

Plus, globally:

Softbank

Examining the HTML shows use of the Chartbeat JavaScript library for visualization.

Nicolas Rapp also created for Fortune 500 this graphic:

Hans Roling

Since 2007 at http://www.gapminder.org/videos/ Hans Roling, a professor of health statistics in Sweden, is an internet legend for his “Joy of Stats” video shown on BBC Nov 10, 2010 and Ted Talks. In them he shows his Gapminder web app which presents multiple dimensions dynamically over time (nearly 300).

Jeffrey Shaffer (@HighVizAbility, co-author of bigbookofdashboards.com) created a time lapse video of his viz showing a trail of dots which grow in size and get darker over time (as the legend notes):

tableau time lapse gapminder

Here, different colors represent different countries, with the United States in red. Most other countries saw a decrease in fertility rate over time while life expectantcy increased.

Within the wonderful Tableau gallery Andy Cotgreave (@acotgreave, now Technical Evangelist at Tableau) built over the years is this re-creation of Gapminder:

tableau gapminder

Andy explained in 2010 how he created the above using Tableau Trendalyzer v6. Download his viz from the Tableau Public website to manipulate using Tableau Public client installed on your laptop.

The trails is a recreation of (@moritz_stefaner) “Remixing Rosling” in @tableau.

Cycle Time

tableau interactive wait times 1956x1372
Click for full pop-up

Metrics include:

Percent of value-add time vs. total time. For example: In the case of physician visits, time interacting with physician vs. waiting and other activities.

H20.ai

h20.ai provides a web-based (SaaS and on-premise) tool that automatically visualizes a time series dataset: VIDEO:

H2o’s software is called “driverless” because it automatically recognizes multiple groups (annual, monthly, weekly, daily, hourly, etc.) by analyzing calculated lags, standard deviations, and other descriptive statistics. It also looks at interactions between trends.

Visit the h2o Aquarium: “Introduction to Driverless AI” Lab

https://www.h2o.ai/try-driverless-ai/

Time Series databases

Introduction to Time Series

KDB proprietary runs in-memory within a single machine.
Graphite - legacy invented by Expedia - does automatic roll-ups of data (losing irregular values)
RiakTS from Basho is built for scale
Promethius for DevOps (highly available with scaleout?)
Cortex is a time-series built on Prometheus to be horizontally scalable, highly available, multi-tenant, long-term storage.
OpenTSDB is a layer on top of HBase.
TimescaleDB, an open source time-series database engineered from PostgreSQL,
InfluxDB - my notes, open-source (MIT)
- No external dependencies (written in Go)
- SQL-like query language
- Input data “Line Format” (not JSON)
- Stores data in compressed format
- Horizontally scaleable (across several servers)
- Kapacitor collects anomalies
Microsoft Azure introduced in 2017 their Time Series Insights.

You need to add your account to it so you can view the Time Series portal at https://insights.timeseries.azure.com for an Environment you define.

InfluxDB Performance Tuning & Schema Design - June 2016 by Gunnar Aasen

Amazon Timestream

More on front-end styling

This is one of several topics:

Wilson Mar