Return to Video

Kurts Favorite ML Algorithm - Intro to Data Science

  • 0:00 - 0:05
    Sure, sure. That's a great question. So I would say, you
  • 0:05 - 0:07
    know, just as a, as a caveat, that really depends a
  • 0:07 - 0:09
    lot of the type of problems any given data scientists are
  • 0:09 - 0:13
    working on, the type of questions you're, you're looking at. For me
  • 0:13 - 0:15
    personally, a lot of the questions that, that I look at
  • 0:15 - 0:18
    and a lot of the things that interest me, tend to
  • 0:18 - 0:23
    be fairly high level and somewhat loosely defined and I mentioned,
  • 0:23 - 0:25
    looking a lot more at the social aspects of how people use
  • 0:25 - 0:29
    social networks. And a lot of topics like
  • 0:29 - 0:31
    that, sometimes the most useful thing is really
  • 0:31 - 0:34
    to just get a broad and qualitative understanding
  • 0:34 - 0:37
    of, of the data you're looking at. So,
  • 0:37 - 0:39
    two of the most valuable techniques I found
  • 0:39 - 0:41
    for that are clustering. There are various approaches,
  • 0:41 - 0:47
    kamines, clustering, hierarchical clustering. And then other dimensionality
  • 0:47 - 0:50
    reduction techniques or things like principle component analysis, PCA.
  • 0:50 - 0:53
    One, one thing, just speaking very specifically that
  • 0:53 - 0:55
    I found very useful is the combination of
  • 0:55 - 0:58
    kamines, clustering and PCA. So if, if you
  • 0:58 - 1:00
    generate clusters and then use PCA to take the
  • 1:00 - 1:04
    most, meaningful vectors to plot those clusters on,
  • 1:04 - 1:06
    you can often do a really nice job of
  • 1:06 - 1:09
    reducing dimensionality of a data set and understanding,
  • 1:09 - 1:12
    the significant differences between clusters in a visual way.
タイトル:
Kurts Favorite ML Algorithm - Intro to Data Science
Video Language:
English
Team:
Udacity
プロジェクト:
ud359: Intro to Data Science
Duration:
01:13

English subtitles

改訂 Compare revisions