search

trinker / termco

Regular Expression Counts of Terms and Substrings
Other
25 stars 5 forks source link

Update validated data set to represent new changes #46

Closed trinker closed 7 years ago

trinker commented 7 years ago

validate_model no gives n.tags and n.classified

trinker commented 7 years ago 
validated <- termco::validated
data.table::setnames(attributes(validated)$tag.counts, 'n', 'n.tagged')

nclass <- data.table::copy(attributes(validated)$tag.counts)
data.table::setnames(nclass, 'n.tagged', 'n.classified')
nclass[, n.classified := c(4, 122, 7, 189)]

attributes(validated)$classified.counts <- nclass

validated
summary(validated)
plot(validated)

And update readme

trinker commented 7 years ago

response_cries sampled is larger than n.classified which is not possible.

-------
Overall:
-------
   accuracy n.tagged n.classified sampled  se lower upper
1:    59.6%      484          322      57 .06 46.9% 72.4%

---------------
Individual Tags:
---------------
              tag accuracy n.tagged n.classified sampled  se lower  upper
1:  back_channels    83.3%        7            4       6 .15 53.5% 100.0%
2: response_cries    72.7%       13            7      11 .13 46.4%  99.0%
3:  justification    55.0%      155          122      20 .11 33.2%  76.8%
4:        summons    50.0%      309          189      20 .11 28.1%  71.9%