3 median smoothing

[hurdles]

Can someone please explain the 3 median line smoothing method.

[Yendall]

Median Smoothing involves finding the median of a certain number of points.

For example:
You have three points
To find the median, you must put these points in order
The median of this data set is

Say you were given a table of values and were asked to perform a smooth:

Value

76.8

76.5

76.1

76.6

76.2

76.0

'3-Median Smoothing':
Find the median of the first 3 values
Put them in order
Median of this set =

'4-Median Smoothing':
You'd then have to find the median of the first 4 values
put them in order

Say you had a table with Years and Sales like so:

Year	Sales
1991	560
1992	680
1993	720
1994	760
1995	580

And you were asked to perform a '3-Median Smooth' for year 1994 you would:

• Identify the sales one before 1994. So 1993 = 720
• Identify the sales one after 1994. So 1995 = 580

Then from this you have your three data points:

• 1993 = 720
• 1994 = 760
• 1995 = 580

From here you have

To perform the smooth, you put them in order
Your median of this set will be

[Stick]

You only use this for non-seasonal time series data, by the way. The most common number groups for median smoothing are 3 and 5 (the weird groupings tend to pop up for mean smoothing).

[Yendall]

You only use this for non-seasonal time series data, by the way. The most common number groups for median smoothing are 3 and 5 (the weird groupings tend to pop up for mean smoothing).

You'd also have to center the data after smoothing it when dealing with smooths that aren't 3 or 5.

[brenden]

Hahaha had my exam 2 prac exam today which I elected to do without a summary book, so I get to a graph in Core and it says "plot using 3-median smoothing" and I fully was like "Shit!! How do I do this! Where do I put the middle one?!" and two seconds later I wanted to smash my face against the table LOL. I also spent a good 15 minutes applying y=a+bx and plugged in the dependent variable value for x and got frustrated that my calc was nowhere near my estimate. Good times.