Minghua Sun

datavizstudio

My First Notebook

Markdown Cheat Sheet

My Second Notebook

Basic Javascript Concepts for Data

Rect VS Bar

Dataviz Types and Their Functions

Isotype / Pictogram Demo

Person Isotype Matrix

Colors, Labels and Interactivity

Data Maps

Plot: labelAnchor

The Art of Data Visualization (Fall 2023)

Let's create a scatter plot of co2 emission (y) vs gdp (x) of 217 nations, colored by life expectancy (quantitative).

Let's create a scatter plot of life expectancy (y) vs gdp (x) of 217 nations, colored by co2 (quantitative).

Is a quantile color scale similar to a log scale?

They solve different problems.

- Log scale transforms the numeric axis to handle wide ranges or exponential patterns.
- Quantile scale transforms how data is grouped or colored by putting equal numbers of data points into each bin.

So:
Log = mathematical transformation of values
Quantile = statistical grouping by distribution

They are conceptually different, but both help when raw data is extreme outliers, skewed or unevenly distributed.

A quantile is a way of dividing your data into equally sized groups.

If you split data into 4 groups, those are quartiles.
Into 10 groups → deciles.
Into 100 groups → percentiles.

Each group contains the same number of data points, even if the numeric ranges differ.

Let's create a scatter plot of life expectancy (y) vs gdp (x) of 217 nations, colored by region (qualitative).

This does:

Start at the last index in the group (latest year).
Look up the value of interest (income, population, etc.).
If it is not null/undefined, return it.
Otherwise, move backward.

Because time-series data often has gaps:

Latest year might be missing income.
Or population missing for 2020, but available for 2019.
Or life expectancy missing for the last few years.

We want the closest valid value up to that year. This makes animations and time sliders smooth.

groupZ, set z scale is "name" to group by country name