top of page

Linear Regression

In order to examine cancer trends over time, we performed least-squares linear regression analysis on various series within our data. Most significantly, we examined trends over time for different states, regions and between these two characteristics. 

Line graph of average cancer rate over time between regions

cancer rate of regions over time.PNG
This graph displays the average cancer rate for each region over time. Here, the cancer rates of all the states in each region were averaged, highlighting how regions different from each other in terms of cancer rates. As is clearly seen, the cancer rates generally decreased over time, but the East and Central regions had less of a decrease than the other three regions, creating a separation in cancer rates between these two groups.

Heatmap of cancer rate correlation over time between states

heatmap of cancer correlations between s

State

State

Here, the cancer rate of each state over time is compared to every other states, displaying the linear relationship between the cancer rate for individual pairs of states. This heatmap shows that most state pairs had somewhat linearly correlated trends in their cancer rates over time. However, some states had very few strong correlations with other states, displayed by the vertical lines in the plot with largely red coloring. Interestingly, states AK through CT were very correlated with each other, possibly due to a confounding characteristic in policy or industry that these states share.

Heatmap of cancer rate correlation over time between regions

Region

Region

As seen above, the correlation was rather high between most regions but two regions, Alaska/Hawaii and the Central region were the least correlated with the rest, indicated their cancer rates didn't vary in similar linear ways over time compared to the other regions.

In this plot, the average cancer rate is plotted along against the average chemical release estimate over time for each state. Further, the timezone regions are are colored to highlight differences between the various regions. Interestingly, the East region had higher cancer rates than other regions, and the Mountain region generally had lower cancer rates. Further, Eastern states also had lower per-capital chemical release estimates compared to Mountain states with higher values. This may be due to the highly urban nature of the US east coast versus the more rural and agricultural characteristics of Mountain states along with population differences between these regions.

bottom of page