THE TOP 100 CHART BUILDER
These pages are experimental -- I'm still learning SQL! These pages may evolve, be
integrated into other pages, etc. They are far from perfect, and all suggestions
are welcome.
WHAT'S IT ALL ABOUT?
These pages attempt to rank the singles based on their chart activity across different
surveys — by Chart date, Radio Station, or Market. You can use the search filters
to create your custom charts, such as "The Top 100 of 1975 for Tucson".
FILTERING
You can add filter values to the text fields at the top of the page. Leave a field
blank to remove filtering on that field. In the date field, you can specify a particular
date or year. Valid values would be "1969-07-31", "1971-05", or even "1967-04-2".
You can also enter a station filter by call letters. Entering "KTKT" will limit your chart
to only songs that appear on KTKT surveys in the ARSA database. An interesting trick: if
you enter "K" in this field, you will limit your search to the so-called western US stations.
Same logic applies, evidently for "W", "X", and "C"... :-)
The Market field allows you to limit your chart data to those surveys for a particular
city, state or country. Mind you, some of the radio stations may be listed under their
city of license. For instance, KCAL is listed as "Redlands, California", whereas you
may have expected to find it San Bernardino.
CAVEAT DUMPSTER
It is important to note the following!!!
- The results are based on the availability of chart data. If you search
a year or market for which the chart data is scant, you will not get very significant
results. Once our ARSA database numbers several thousand surveys per year, you will
be able to use these chart results with confidence.
- These pages only take into account those songs ranked by position in their
respective surveys. A survey may list a major hit, but if the survey didn't assign
a chart position to that hit, it won't be considered in the generated charts.
- The algorithm I currently use for ranking the singles is probably not the best
and certainly not the definitive for ARSA. As more chart data is available, as more
chart weeks are available, I will revise the algorithms accordingly.
- Further to the previous note, the use of "LW" data in ranking (discussed below)
is problematic in that it potentially doubles rankings should ARSA contain two charts
for consecutive weeks.
- You should currently consider the "POINTS" column, as I have not yet determined
a satisfactory way of generating unique chart positions, i.e. 1, 2, 3, etc.
(As I said, I'm still learning SQL...)
THE RANKING SYSTEM
Songs are first compiled based on the search filters provided, if any. The selected
songs are then awarded points based on the following system:
- Chart position. For each chart containing the song, points are given based on
the song position (1=1000 points, 2=500 points, 3=333, 4=250, 40=25, 50=20 points, 100=10 points)
- Additional bonus points are added in for chart position (1-5 => 20 points, 6-10 => 10 points,
11-20 => 5 points, 21-30 => 2 points)
- Because the ARSA chart data is still scant, we use LW chart data whenever available
to relativize the data. If the song was number one last week, 100 points are added; 80 points
are added for number two; else if the song maintained a top five position both weeks, 40 points
are added in. Finally, if the song is moving up the charts, we factor in difference between
last and this week's positions, limited to 20 points for a fast chart climber. For instance,
if the song was 23 last week and 4 this week, we add in 19 points.
- The number of surveys charting a song is a good indication of its popularity. Currently
we add in the square of the total number of surveys containing a song. If the song appears
in three charts, we add in 9 points. If the song appears in 5 charts, we add in 25 points.
- Finally, if the chart includes "Weeks On" data, we add in those weeks.
Complicated? A bit. Accurate? Not likely. Interesting browse? Hopefully. Anyway,
it's a start.
<***>