Permalink
Commits on Mar 22, 2016
  1. Merge pull request #11 from modest/hide-ignored-checkbox

    Usability: Hide checkbox when irrelevant
    jmikkola committed Mar 22, 2016
Commits on Mar 16, 2016
  1. Usability: Hide checkbox when irrelevant

    The "Use multiple testing correction" checkbox is visible and enabled by default. Confusingly, this checkbox has no effect at all unless the user clicks "Add another group" to append 2nd variation.
    
    This change hides the checkbox in scenarios where it does not affect the calculation (<=1 variation).
    modest committed Mar 16, 2016
Commits on Nov 5, 2015
Commits on Nov 3, 2015
Commits on May 18, 2015
  1. Merge pull request #9 from blubber/python3

    Python: fix Python 3 compatibility
    
    Thanks Tiemo!
    gostevehoward committed May 18, 2015
Commits on May 11, 2015
  1. Python: fix Python 3 compatibility

    Added `sys` import and bound `xrange` to `range` in case in case the
    Python version in use is version 3 or greater.
    blubber committed May 11, 2015
Commits on Jan 7, 2015
  1. simplify InputsView interface, removing addInputRow

    Presenter can accomplish the job with just InputsView.{get,set}Inputs(), which also allows
    (slightly) more logic to be under test!
    gostevehoward committed Jan 7, 2015
  2. Permit removing the baseline input row

    Fixes #7, which rightly points out that this can be an annoying
    restriction.
    
    Rather than permit the UI to show no input rows when removing a lone
    baseline group, this implementation simply clears the baseline row's
    inputs when there are no other groups. One could also hide the remove
    link in that situation, but this seems more consistent and convenient.
    
    For the implemenation, I delegated all the meaty logic to Presenter
    using a new callback on InputsView. Presenter accomplishes the task
    just using InputsView.{get,set}Inputs(). I realized adding a group
    could work this way too, simplying the interface between the two
    objects and putting more logic under test. Refactoring coming...
    gostevehoward committed Jan 7, 2015
Commits on Aug 22, 2014
Commits on Jun 3, 2014
Commits on Aug 13, 2013
Commits on Oct 12, 2012
  1. Merge pull request #4 from visualmotive/master

    Script for Mixpanel A/B test integration
    showard committed Oct 12, 2012
  2. Move Mixpanel script into a tools directory

    Chris Mueller committed Oct 12, 2012
  3. Add a simple Mixpanel integration for A/B tests

    Chris Mueller committed Oct 12, 2012
Commits on Sep 28, 2012
Commits on Aug 3, 2012
  1. Merge pull request #3 from rishibajekal/master

    Fix "NaN%" and syntax
    showard committed Aug 3, 2012
Commits on Aug 1, 2012
  1. Add more meaningful variable names

    Rishi Bajekal committed Aug 1, 2012
Commits on Jul 31, 2012
  1. Display 'N/A' instead of 'NaN%' and fix some syntax

    Rishi Bajekal committed Jul 31, 2012
Commits on Jun 18, 2012
  1. add UI options for interval conf level and mult test correction

    - confidence level of all confidence intervals was hardcoded to
      95%. now you can enter any confidence level, which is useful if
      you're making decisions based on the confidence intervals.
    - can now disable multiple testing correction if you feel it's not
      applicable or otherwise just want to see the effect of disabling it.
    showard committed Jun 16, 2012
Commits on May 7, 2012
Commits on Apr 12, 2012
Commits on Mar 21, 2012
  1. use the same confidence level for all confidence intervals

    it's more logical, even though it leads to the slightly confusing UI
    situation where the success rate intervals clearly overlap but the
    improvement interval does not include zero.
    showard committed Mar 20, 2012
Commits on Mar 20, 2012
  1. show CI upper/lower limits and actual (not smoothed) estimators

    - Render CI limits rather than the center of the internal plus/minus
      the half-width. This is more directly useful and more in line with
      what's normally done.
    - Show ML estimators for success rate and improvement rather than
      smoothed estimates. While I think they've valuable, the smoothed
      values are unusual and confusing, and it's probably better to just
      use the interval.
    - Put the interval front and center and de-emphasizes the estimated
      value, since the interval is more statistically meaningful.
    showard committed Mar 20, 2012
  2. don't format p-values as percentages

    p-values don't normally appear as percentages, raw values are the
    common practice.
    showard committed Mar 20, 2012
Commits on Mar 16, 2012
  1. Simplify the demo page

    Chris Mueller committed Mar 16, 2012
  2. Use better SVG detection

    Chris Mueller committed Mar 16, 2012
Commits on Mar 15, 2012
  1. Do not render SVG in IE

    Chris Mueller committed Mar 15, 2012
  2. Fix some trailing commas for IE

    Chris Mueller committed Mar 15, 2012