GEOG 371 Lecture Notes - Lecture 13: Scatter Plot
Week 13
4/9 – Rank Correlation
• Outliers – oservatios that dot fall o the tred
• How to decide which transformation to use (e.g. log10, ln, square root, etc.)
o Look at scatterplot
o Trial and error – Try different transformations and choose the transformation that gives
the largest r value
o Trasfor data to ahieve liear assoiatio: = f ad/ or = f
▪ X = log10(x)
▪ Y = log10(y)
o 1st – transform x and/ or y
o 2nd – redo scatterplot w/ transformed variables
o 3rd – check to see if its linear
▪ If its not, try another transformation
Rank Correlation – Spearas ‘ ‘s
• Use when one or more of these exist –
o Ordinal data
o Have a small sample size, n < 20
o Data are poorl distriuted
▪ Big outliers
• Spearas ‘:
•
o R(Xi) = Rank of obs i on variable X
o R(Yi) = Rank of obs i on variable Y
o di = R(Xi) – R(Yi)
o n = # observations
• Ex –
o Is preference related to winning %?
Team
Preference Rank
(x)
Winning % (y)
Ri(Y) – ranking
based on wins
di (x rank – y
rank)
Cardinals
1
0.531
3
-2
Indians
3
0.591
4
-1
Yankees
2
0.518
2
0
)( 1
6
12
1
2
nn
d
r
n
i
i
s
find more resources at oneclass.com
find more resources at oneclass.com