Basic HTML Version
© 2014 Pacific Crest
343
●
Deciding the correlation is casual when it is not
Example
:
Global Average Temperature vs. Number of Pirates
Global Average Temperature, °C
Number of Pirates (Approximate)
35000 45000 20000 15000 5000 400 17
16.5
16.0
15.5
15.0
14.5
14.0
13.5
13.0
2000
1980
1940
1920
1880
1860
1820
“PiratesVsTemp(en)” by PiratesVsTemp.svg:
RedAndrderivative work: Mikhail Ryazanov (talk)
- PiratesVsTemp.svg. Licensed under Creative
Commons Attribution-Share Alike 3.0-2.5-2.0-1.0 via
Wikimedia Commons (link available online).
Why?
Although the relationship looks linear and the number of pirates has decreased
consistently while the global average temperature increased consistently, does anyone
really believe that permitting pirating is a solution to global warming concerns?
●
Illogical independent/dependent assignments
Example
: Instead of trying to predict scores on the test, we want to predict scores on the quiz.
Why?
Some variables are sequential with time, space, or steps in a process. Thus the first
event must be the independent variable, while to second event must be the dependent.
●
Predicting for a variable outside the range of measured values
Example
: Ice cream sales and temperature, when the temperature is above 25 degrees centigrade.
Sales
Temperature °C
$0
$100
$200
$300
$400
$500
$600
$700
10
15
20
25
Why?
The relationship is strong between 10 and 25 degrees but breaks down and changes
considerably beyond 25 degrees, as we’ve already seen. Thus, a prediction made
for 40 degrees, for example, will be well off the mark, based upon the results for
temperatures between 0 and 25 degrees.
7.4 Simple Linear Regression