PlayOn: August 2005 Archives

« July 2005 | Main | September 2005 »

August 31, 2005

Change in Leveling Time Over Time

Server Sample: RP (High), PvE (Medium), PvE (High), PvP (High), PvP (High)
Sampling Period: 8/07/2005 12:00 am - 8/14/2005 12:00 am
Sampling Resolution: ~12 minutes
Parsing Method: The sample unit is leveling event. We tabulate the time between a character's level and when we observe them at a new level. Only a player's online time is counted. We exclude the first leveling event from every character because it doesn't constitute the total amount of time to make that level.
Data Filter: None
Sample Size: 149,240 leveling events

We reported before that on average it takes characters about 20 days to get from level 1 to 60. While analyzing newer data for other purposes, we noticed that the average leveling time had decreased. We then double-checked our numbers to make sure we weren't looking at a parsing or collation error. While our double-checking showed that the average leveling-time had indeed decreased, we uncovered an error made early on in estimating our sampling resolution.

The actual sampling resolution for the previous data was actually lower than we had estimated (every 11.82 minutes instead of 13.50 minutes - we have since switched to an algorithm that calculates exact time between samples). So the average leveling time to get to level 60 from our previous data set (early June 2005) should have been 18.0. The average leveling time to get to level 60 from our current August data set is 15.3. That's about a 15% reduction in leveling time. Our hunch is that we're seeing the activity of alt characters lowering the overall leveling time.

Posted by Nick

Posted by nickyee at 11:00 AM | Comments (5) | TrackBack

August 30, 2005

Zone Population Map

Server Sample: RP (High), PvE (Medium), PvE (High), PvP (High)
Sampling Period: 7/05/2005 12:00 am - 7/12/2005 12:00 am
Sampling Resolution: ~14 minutes
Parsing Method: We parsed the number of characters in each zone. Characters just passing through (seen for only one snapshot and not the next are not counted). Characters who were in 2 zones in an hour had both zones counted.
Data Filter: None
Sample Size: 50,464 zone counts

We used the zone population data and a bit of Photoshop to create the zone population map below.

Posted by Nick & Nic

Posted by nickyee at 02:35 PM | Comments (6) | TrackBack

August 29, 2005

Zone Populations by Character Level, Zone Type, and Time of Day

Server Sample: RP (High), PvE (Medium), PvE (High), PvP (High)
Sampling Period: 7/05/2005 12:00 am - 7/12/2005 12:00 am
Sampling Resolution: ~12 minutes
Parsing Method: For each one hour period, we parsed the location of every unique character. Zones that a character is only observed once in (traveling through) are not counted. For each character per hour, only the first zone they are seen in is counted.
Data Filter: None
Sample Size: 1,249,017 Counts

We split character levels up into 3 bands: 1-20, 21-40, and 41-60. We then plotted the average zone population over time of day by zone type - Instance, Outdoor, Main City, and PvP (Alterac Valley and Warsong Gulch).

Posted by Nick & Eric

Posted by nickyee at 11:00 AM | Comments (1) | TrackBack

August 26, 2005

Zone Populations by Time of Day

First, we graphed the average population in each zone. Unsurprisingly, Ironforge is the most populated zone, followed by Ogrimmar and The Barrens.

Next, we chart the zone populations by time of day.

Among the main cities, the average zone population (by proportion) is fairly stable over the day.

The same is true for the outdoor zones.

Among the instances, Molten Core sticks out as the only zone that seems to fluctuate proportionately as a function of time. Also, the instance population counts swell more in the evening hours than do in the Outdoor and Main City zones.

Posted by Nick & Eric

Posted by nickyee at 10:57 AM | Comments (7) | TrackBack

August 18, 2005

Guild Abandonment

Server Sample: RP (High), PvE (Medium), PvE (High)
Sampling Period: Week 1: 6/10 - 6/17. Week 2: 07/05 - 07/12.
Sampling Resolution: ~14 minutes
Parsing Method: The sample unit is a guild. We tracked all guilds observed in the June week and counted the number of unique characters observed to belong to each guild. We then did the same for all guilds observed in the July week. We then tabulated whether a guild observed in June was observed again in July.
Data Filter: None
Sample Size: 2,744 guilds.

Of the 2,744 guilds observed in the one-week period in June, 21% were not observed again in the one-week period in July. However, many of these guilds (19%) were only observed with one character in June. If we only take into account guilds with more than one observed character in June, then 13% of those were not observed again in July. If we increase that criteria to guilds with at least 5 characters, then 7% of those were not observed again in July.

Of the guilds that were observed in both June and July (N = 2168), the average change in guild size was exactly 0. In other words, on average, guilds did not grow or shrink over the one month period. This was somewhat surprising given that we might have expected guilds to grow over time. In fact, if we exclude all "one-person guilds", the average change is mildly negative (-.13). On the other hand, the range of change in guild size was quite large even though the overall change was near 0. The largest loss was 103 and the largest gain was 87.

Posted by Nick

Posted by nickyee at 10:10 AM | Comments (0) | TrackBack

August 15, 2005

Predictors of Character Abandonment

Server Sample: RP (High), PvE (Medium), PvE (High)
Sampling Period: Week 1: 6/10 - 6/17. Week 2: 07/05 - 07/12.
Sampling Resolution: ~14 minutes
Parsing Method: We tracked all unique characters in both periods. For each character, we noted their playing time, guild affiliation, group ratio, starting level, class, and race. Then we tabulated which of the characters observed in June were not observed in July (both 1 week periods). This way we generated two lists of characters - those who were observed again after a month and those that were not.
Data Filter: None
Sample Size: 75,314 characters

Once we had the list of characters along with whether they were observed again or not, we ran a logistic regression with the following variables as predictors: playing time, grouping ratio, level, guild size, and guild affiliation.

As expected, level and playing time were the best predictors of abandonment. Whether the character was guilded was the third best predictor. Interestingly, as long as a character was guilded, the size of the guild did not matter.

The model had an r-square of .23 and had a classification success rate of 75% for abandonment and 70% for retention.

Posted by Nick

Posted by nickyee at 02:16 PM | Comments (0) | TrackBack

August 03, 2005

Character Abandonment

Server Sample: RP (High), PvE (Medium), PvE (High)
Sampling Period: Week 1: 6/10 - 6/17. Week 2: 07/05 - 07/12.
Sampling Resolution: ~14 minutes
Parsing Method: We tracked all unique characters in both periods. For each character, we noted their playing time, guild affiliation, group ratio, starting level, class, and race. Then we tabulated which of the characters observed in June were not observed in July (both 1 week periods). This way we generated two lists of characters – those who were observed again after a month and those that were not.
Data Filter: None
Sample Size: 75,314 characters

Of the 75,314 characters observed in the one-week period in June, 46% were not observed in a one-week period in July. The lower level the character, the less likely they were observed again in July.

Now, there are other possible reasons for not observing a character again, such as being away for vacation or cycling through another character. But another way of looking at the data seems to suggest that we are indeed looking mostly at abandonment. Below is the histogram of characters who were observed in July but not in June. In other words, we are looking at a mix of new characters (who began after the June sample) and old characters (who didn't log on during the June sample). The majority of these characters are low level characters (as suggested by the abandonment hypothesis) rather than mid or high level characters (as suggested by the cycling hypothesis).

Interestingly, the server populations grew over this period. The June sample had 75,315 characters. The July sample had 83,253 characters. What is unclear is the relationship between character abandonment and player churn. Once we get more data, we'll rerun this using one month time frames.

Posted by Nick

Posted by nickyee at 11:13 AM | Comments (0) | TrackBack

August 01, 2005

PvP Server Differences (Part 2)

Server Sample: RP (High), PvE (Medium), PvE (High), PvP (High)
Sampling Period: 6/30/2005 2:00 am - 7/07/2005 2:00 am
Sampling Resolution: ~14 minutes
Parsing Method: The sample unit is each unique character. Each character was tracked across the server logs. Total playing time, lowest observed level, highest observed level, guild affiliation, and zones seen in were parsed.
Data Filter: None
Sample Size: 106,762 characters

Here, we continue the PvP server differences findings.

Grouping Ratio

Characters on the PvP server were observed to be in a group more often than characters not on PvP servers. Overall, the difference was about 30% vs. 25%. The increased grouping ratio seems to be reflected across all 60 levels.

Level Progress

We calculated the level progression that each observed character made over the one week period. Characters on PvP servers made a significantly higher, but marginal, difference over the one week period than the other servers.

Leveling Time

Messages on forums seemed to suggest that it was harder and took longer to level on PvP servers than on non-PvP servers. While we did find a significant difference, it was almost in the opposite direction. Characters on the PvP server leveled faster than most of the other servers. In fact, perhaps the need to level is more salient on PvP servers than non-PvP servers and outweighs the difficulty of leveling.

Posted by nickyee at 10:40 AM | Comments (2) | TrackBack