2

Sorry but I'm not a statics expert at all, but I'm following some on line course and it is fascinating me.

I just found the existence of the mode:

The mode is the value that appears most often in a set of data. (Wikipedia)

What about a sequence where the values are different one from each other? Like

l = [1, 1.2, 1.3, 0.9, 12, 5]

I'd like to take out from this sequence a value that is not the average (that is 3.5) because it is not explicative enough so I thought to discretize (like [1, 1, 1, 1, 12, 5]) them and then take the mode (that would be 1) but I'm sure that there is a better way.

nkint
  • 1,823
  • is there a tag for say "basic concept"? – nkint Jul 25 '13 at 21:49
  • Split the data to chunks. How we do it is has no real rules, say $0$ to $2$, $2$ to $4$, $4$ to $6$, and so on. Mode is $1$ (midway point serving as identifier). Not very persuasive, we have hardly any data. For say $100$ data points, the analogous procedure begins to be reasonable. Mode is really not important, though bimodality can be (clusterings around two values. – André Nicolas Jul 25 '13 at 21:57

1 Answers1

1

There's no uniquely Right Way to define the concept that you're grasping at, but you're not alone in wanting to do so. The (ill-defined) problem you're trying to solve is called cluster analysis, and there's an entire field of research dedicated to it.

Daniel Franke
  • 1,204