Skip to content

As a user experience manager I want to understand the types of text users are putting in.  #76

Closed

Description

@vronvali
map: 
function(doc) {
  emit(doc.originalText.length, [new Date(doc.dateCreated),  doc.originalText.length, doc.originalText]);
}
  • Sort text by length.
reduce: 
_count

screen shot 2015-06-05 at 6 36 03 pm

  • Group text by user to see how many texts users make on average.
  • Try to group user by type based on the length of what they write. (Do in a Spreadsheet)

Activity

  1. self-assigned this
    on Jun 5, 2015
  2. vronvali commented on Jun 8, 2015

    @vronvali
    MemberAuthor

    After narrowing down the data on CouchDB to only listing the word count in comparison to users it was easier to see a trend to help identify the amount of each type of user. Before beginning, we hypothesized that less than 5 words was junk, 5-100 words was tag cloud users, 100-300 were made by word cloud users, and above 300 were made by language cloud users.

    After looking at the data. I decided that each dramatic change in cloud count symbolized the different categories. This hypothesis led to finding out that anything less than 7 words was junk, 7-11 words was tag cloud, 12-99 were word clouds, 43-99 were a new category I called Experienced Word Cloud Users(EWCU), and 100 and up was iLanguage Cloud users. I made these assumptions because of as the amount of words went up the amount of people drastically went down.

    screen shot 2015-06-08 at 2 37 48 pm

    Which can be seen by the averages and Standard Deviations of each group. My assumption is most people want to use this app as a tag cloud, thus why there are less words, but more clouds in this category. However, after reading the Standard Deviation graph, it can be assumed that most people are consistently using word cloud over any other type. detecting word clouds with a significant degree confidence is not really possible since there is a large standard deviation on the word cloud category.
    screen shot 2015-06-08 at 2 37 59 pm

  3. cesine commented on Jun 8, 2015

    @cesine
    Contributor

    @vronvali i think the standard deviation chart is best viewed as a box plot, have you seen those before?
    example:

    https://developers.google.com/chart/interactive/docs/gallery/candlestickchart
    screen shot 2015-06-08 at 2 43 03 pm

    we can move on to the tests now that we have an idea of what we want for categories, and do a box plot tutorial FieldDB/FieldDB#2008
    next time you have to do histograms

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions