Comparing two programs for same data
- From: frederik_coppens@xxxxxxxxx
- Date: Tue, 28 Aug 2007 01:31:18 -0700
Hi,
I have the following problem. I have a text corpus in which each word
is tagged with a semantic concept. I have two different algorithms
that both try to assign the correct tag to each word. After running
each algorithm, I can measure the performance of each algorithm by
looking at the number of tokens that it tagged correctly.
For example, suppose there are 3000 words, algorithm1 tags 2115 of
them correctly, and algorithm2 tags 2200 of them correctly.
Now my question is: if one of the algorithms does better than the
other, which test can I apply to check if the difference is
statistically significant?
My knowledge of statistics is superficial. For an expert, this
question is probably not too difficult. Thanks in advance,
Frederik
.
- Follow-Ups:
- Re: Comparing two programs for same data
- From: Richard Ulrich
- Re: Comparing two programs for same data
- From: Phil Holman
- Re: Comparing two programs for same data
- Prev by Date: Re: Forecasting with mixed variable
- Next by Date: Re: Forecasting with mixed variable
- Previous by thread: Forecasting with mixed variable
- Next by thread: Re: Comparing two programs for same data
- Index(es):
Relevant Pages
|
