I am trying to find correlation between Opioid Severity and Employment Statistics. I don't have access to my professor right now.

So I have 2 datasets. 1 dataset measures Opioid concentration (Positivity Rate) in various regions of a state per year. Call this opioid

Year   #Regions  PositivityRate  State
2007    5        0               CA
2007    2        1               CA
2007    3        2               CA
2007    4        3               CA
2008
.
2009
.
.
2014

So, the a high number in PositivityRate means that the there is MORE concentrated and the severity is high! #Regions simply is referring to how many regions in the state of CA have that amount of PositivityRate

Now I have a second dataset which has the employment statistics of CA: call this employCA

Year   Month  Labor_Force  Unemployment_Rate  Current_Employed   Current_Unemployed

2007    Jan    32423       4%                 21341              12343
2007    Feb    32423       4.1%               21234              12213
.
.
.
2007    Dec    12423       1%                 12433              23452
2008    Jan
.
.
2009
.
.
2014

I made those numbers up, but you get the idea.. I have data for 12 months, for years from 2007 to 2014

Now what I'm doing is, grouping opioid by Year and filtering for state == CA. and finding the weighted average to represent severity

So, for the dataset opioid, For the state == CA, and year == 2007, the severity would be

$\frac{(5 \times 0) + (2 \times 1) + (3 \times 2) + (4 \times 3)}{5 + 2 + 3 + 4}$, Now in the employCA, I group by year and find the mean of Labor_Force, Unemployment_Rate, Current_Employed, Current_Unemployed

So my final two tables would look like

opioid:

Year    Severity     State
2007        2         CA
2008        3         CA
2009        3         CA
2010        3         CA
2011        1         CA
2012        3         CA
2013        4         CA
2014        2         CA

and the employCA dataset now has

Year     MeanLabor   MeanUnemployRate   MeanEmployed    MeanUnemployed 
2007
2008
2009
.
.
.
2014

Now I join the two tables on Year and compute Pearson correlation between the Severity and the remaining columns excluding Year

My questions are:

  1. Am I measuring the Opioid Severity the right way? Using weighted average?

  2. Is it right to aggregate using average of various columns?

  3. The data is not scaled.. Is it okay to compute correlations between variables considering that all the data are in different scales? Does scaling the data improve what the correlations represent? If I'm supposed to scale it, could you point to me how I do it?

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.