Mussel Watch

Written: June 2018 (Kernel)


Here is a summary of the analysis I performed on the Mussel Watch dataset. If you would like to see the complete analysis, please see my kernel on Kaggle.


Mussel Watch is a marine monitoring program under NOAA founded in 1986. The program monitors more than 140 contaminants across the major waters in the United States.

Time Scale

Here's what the yearly records look like, the data covers 30 years from 1986 to 2015. Approximately 50% of entries are before 2001 and 50% of entries are after 2001.

Figure 1


Figure 2 shows the most popular places where Mussel Watch operates.

Figure 2

Contaminants in Organisms

I'm going to split the data into two time periods... Data recorded before 2001 and data collected on 2001 or later. I'd like to see if contiminant levels have changed in anyway over the past 30 years.

If we look at the plot below, it appears they have! Almost all the substance levels dropped earlier levels. This plot gives me hope that new environmental practices are helping.

Figure 3: The substance levels found in different organisms.

Contaminants Over Time

Figure 4 shows the median measurement levels for a substance in a particular year. From looking at figure 2, we can see different organisms have different substance levels. To avoid variation as a result of different organisms, I restricted the following plot to only sediment measurements.

Figure 4: The substance levels found in Sediment measurements over time.

Mussel Attributes

Finally, we can see below the different shapes and sizes of mussels.

A few things to note:

  • The relatioship between width and length looks approximately cubric
  • The smallest bivalves are from the Great Lakes. I imagine this is because rivers and the ocean have more oxygen than a still lake and can thus support bigger organisms
  • "Alabama to Branzos River" has the biggest spread and heaviest organisms
  • "Florida Gulf Coast" has a simliar distribution to "Florida Keys to Cape Hatteras"

Figure 5: The relationship between length and width.


Thank you for viewing this summary, if you would like to see the complete, more extensive version, please see my kernel on Kaggle.