JavaScript is required to use Bungie.net

Forums

originally posted in: If you have a 1.0 kdr...
1/13/2016 11:39:22 AM
2
You're averaging averages there and that doesn't work as it becomes weighted. You have to tally up total kills and total deaths then do your K/D ratio. Which would make a K/D of 1. (I'm an analyst).
English

Posting in language:

 

Play nice. Take a minute to review our Code of Conduct before submitting your post. Cancel Edit Create Fireteam Post

  • And since we are sharing credentials, I have 2 engineering degrees and a math degree.

    Posting in language:

     

    Play nice. Take a minute to review our Code of Conduct before submitting your post. Cancel Edit Create Fireteam Post

  • Since you are an analyst (of what, I don't know), let's get technical. The term "average K/D" is ambiguous since there are different kinds of averages (in common parlance, average typically refers to an arithmetic mean). If you take the total number of kills and divide it by the total number of deaths that is, technically an average. But it is useless as a metric of player performance, because it does not in any way include the per-player distribution (i.e. player performance). It is a measure of the expected number of kills generated by each death. It is like trying to represent duck hunter performance by calculating the average number of dead ducks per dead duck and then saying anyone who kills more than 1 duck is above average. They are 2 different metrics (ducks/hunter vs ducks/duck). In a similar example, consider a group of people with cars. If you take the total miles driven by all of them, and divide by the total amount of gas used by all of them, you get an average MPG. If you calculate the individual MPG of each car and then the mean of those values, you get another (different) average MPG. Which one would you use to determine if one person's car was above average at fuel conservation? Not the first number, which ignores the cars entirely to determine the mileage produced by an average gallon of gas consumed. The latter number is a per-vehicle representation of fuel consumption. All kills divided by all deaths is a metric of how many kills a single death is expected to result in, but has nothing to do with player performance. The mean player K/D, on the other hand, is a metric of typical performance of an individual player.

    Posting in language:

     

    Play nice. Take a minute to review our Code of Conduct before submitting your post. Cancel Edit Create Fireteam Post

  • I'm a trade/commerce analyst that works mainly with SQLs and Excel. You're taking averages (in this case ratios) and using them like whole numbers. It doesn't work out like that. What your doing is like this small example (using %s here, as they are another form of average). In Ireland, 90% of gamers play Destiny In the UK, 90% of gamers play Destiny In China, 5% of gamers play Destiny According to your method the total of players of these 3 countries that play Destiny would equal 61.6%

    Posting in language:

     

    Play nice. Take a minute to review our Code of Conduct before submitting your post. Cancel Edit Create Fireteam Post

  • No, what I am saying is that a country in which 61% of players play Destiny is about average for a country, when countries are ranked by Destiny adoption rate. My point being that each average has a specific meaning in a specific context. Defining a metric of player performance and comparing individuals' metrics to the average of all players' metrics is valid *in that context.* The fact that the metric is an average of something else is irrelevant. The total number of kills divided by the total number of deaths is an average, but it is not a determination of an average player. It is a determination of an average death weighted by kills. "The average *death* is associated with one kill." When you take a per-player metric, regardless of what that metric is, and average it over all players you get a metric for an average player. It might be average number of hours played *per player*, average number of crucible matches *per player*, average number of emotes used per raid *per player*, average DTR score *per player*, or average K/D per player. Any of those represents an average *player* when considering a specific metric. If the average player uses 3 emotes per raid and I use 300 emotes in a KF run, my emote usage was above average. The fact that "emotes used per raid" is itself an average is irrelevant to that fact. (Whether any metric is or is not a measure of player skill is a separate issue.)

    Posting in language:

     

    Play nice. Take a minute to review our Code of Conduct before submitting your post. Cancel Edit Create Fireteam Post

  • and this is why K/D means Deaths. Its still a good way to "Judge" a players skill but its their skill at not being killed....lol

    Posting in language:

     

    Play nice. Take a minute to review our Code of Conduct before submitting your post. Cancel Edit Create Fireteam Post

  • What he said!^

    Posting in language:

     

    Play nice. Take a minute to review our Code of Conduct before submitting your post. Cancel Edit Create Fireteam Post

  • #Rekt2016

    Posting in language:

     

    Play nice. Take a minute to review our Code of Conduct before submitting your post. Cancel Edit Create Fireteam Post

You are not allowed to view this content.
;
preload icon
preload icon
preload icon