Return to University

Coursera – Data visualization

After the fun experience of participating in the EdX course Introduction into Computer Science I opted for another course, this time on Coursera. Data Visualization by Prof. John C. Hart – so far it is also quite fun.

The homework is peer-graded and because of the honor-code I cannot publish source code that generated the images.
I used Python 2.7 with NumPy and Matplotlib to analyze and plot the data. I think these open-source tools are really worth learning and for most tasks equal or superior to multi-thousand-dollar software.

First data set

The first data set analyzes the Land-Sea temperature index between 1880 and 2015 for each month.
Observations for the plot below:

  • The data (independent data = time)for x-axis is continuous so a line graph is appropriate
  • I chose Months as dependent data in categories, so I opted for a stacked stream graph.
  • Since all 12 moths are too much and the individual months do not show significant differences I selected 4 months, one for each season
  • I scaled the data from .01 degrees Centigrade to 1 Degree centigrade, otherwise the values on the y-axis would be quite confusing
  • Years with any unavailable data (only 1880 and 2015) have been dropped before the plot
  • The colors are chosen for maximum readability – admittedly it could be more aesthetic
  • I plotted the same data three times playing around with the base-line. I think it is nice to see the increase starting 1960

Click on the plot for a full-scale image.

Land-sea temperature index for selected months between 1880-2015

land-sea temperature index 1880-2015 for selected months

Second data set

The second data set analyzes the same Land-Sea temperature index between 1880 and 2015 but now with focus on geographical zones.
Observations for the plot below:

  • The data (independet data = time)for x-axis is continuous so a line graph is appropriate
  • As dependent data I show selected hemispheres and zones, this time as a line-graph.
  • I scaled the data from .01 degrees Centigrade to 1 Degree centigrade, otherwise the values on the y-axis would be quite confusing
  • First I wanted to show how the land-sea temperature index of two hemispheres evolved over time in comparison to he global mean
  • When I discovered that for the northern hemisphere the land-sea temperature index was higher than on the southern hemisphere I tried to figure out wich zone influenced most.
  • The second plot clearly shows a high volatility for the polar zone and a sigificantly higher value than for the other zones

Click on the plot for a full-scale image.

land-sea temperature index 1880-2015 by climate zones

land-sea temperature index 1880-2015 by climate zones

Leave a Reply