reid@decwrl.dec.com (Brian Reid) (05/16/88)
I thought that people might be amused by the following chart; it summarizes
the last 24 months of arbitron reports.
These numbers differ somewhat from some of the data posted monthly, for
2 reasons. First, because I've used the April 88 algorithms on older
data: I've upgraded my arbitron processing several times through the
years. Second, in the process of preparing this report I found a bug in
the code that estimates the total number of users (not that that number
is particularly reliable anyhow). This is why I keep raw data rather
than digested data.
Anyhow, here it is. HC12, HC3, and HC1 are 3 different ways of
computing the "Host Count". HC12 is the total number of distinct hosts
mentioned in Path lines arriving at decwrl in the last 12 months; HC3
is the total number of distinct hosts mentioned in Path lines arriving
at decwrl in the last 3 months, and HC1 is the total number of distinct
hosts in path lines in the past 30 days. "Sample size" is the number of
sites submitting arbitron reports, "NG count" is the number of
newsgroups carried by decwrl that had non-zero traffic during the
measurement interval. There are lots of explanations for why HC1 and
HC3 and HC12 are so different; rather than bore you with the
explanations I'll just give you the numbers.
month NG msgs Mbyte HC HC HC sample reader user
count count count 12 3 1 size estimate estimate
==============================================================================
apr86 241 28388 59.512 2193 2193 2315 249 53000 380000
may86 244 27476 59.155 2618 2618 2352 296 59000 420000
jun86 243 20706 49.024 2951 2951 1724 293 75000 500000
jul86 236 19521 46.218 3052 2828 1692 320 95000 640000
aug86 241 19057 44.846 3354 2311 1641 351 78000 630000
sep86 262 19652 45.593 3607 2506 1765 320 82000 660000
oct86 298 20579 49.490 3878 2502 1816 328 88000 640000
nov86 314 18259 43.233 4104 2550 1684 354 88000 590000
dec86 242 17396 44.914 4358 2585 1656 356 84000 620000
jan87 252 26809 62.321 4683 2653 2056 347 79000 570000
feb87 250 29410 62.735 5053 2873 2223 362 90000 590000
mar87 259 28223 62.036 5376 3149 2238 398 90000 550000
apr87 - - - 5200 2765 2238 412 86000 570000
may87 259 29689 61.272 5052 2945 2249 435 105000 550000
jun87 269 32349 71.316 5323 3453 2359 452 120000 530000
jul87 262 33493 77.531 5519 3514 2585 476 118000 640000
aug87 265 26689 57.344 5675 3552 2302 497 125000 720000
sep87 312 32665 74.395 5929 3711 2632 514 134000 810000
oct87 319 33161 69.593 6195 3871 2837 534 131000 770000
nov87 321 40843 82.451 6368 4064 2797 589 140000 730000
dec87 328 34839 77.438 6615 4130 2854 596 129000 770000
jan88 330 50593 110.838 6925 4363 3337 645 137000 840000
feb88 - - - 7362 4838 3498 632 143000 850000
mar88 347 53144 121.561 7362 4398 3498 641 142000 910000
apr88 381 57979 130.945 7810 4803 4152 809 141000 880000
The archive tape for apr87 traffic was unreadable, and I had a screwup in my
crontab that caused the feb88 traffic data not to get recorded. Everything
else is real. I now have my own big disk to keep this on, so all the data is
now online.
I sure wish I had more than 2 years of data. I didn't start doing arbitron
until January 86, and I didn't start archiving the data until I came to
Digital from Stanford in April 86.
Brian