101-200.Rmd

```{r, echo=FALSE, purl=FALSE, message=FALSE}
#knitr::opts_chunk$set(results='hide', comment = "#>", purl = FALSE)

library(knitr)
library(kableExtra)
library(stringr)
library(dplyr)

#c_gviz <- read.csv("../../API_tests/catal_sample_1000_GV_output.csv") # read peter's data here
#thumbs <- list.files("thumbs/") # take first 100 > make 10 pages.
#filenames <- str_remove(thumbs, "thumb_")
filenames <- list.files("thumbs/") %>% str_remove("thumb_")

c_gviz_sub <- read.csv("../../API_tests/catal_sample_1000_GV_output.csv") %>%   
    filter(filename %in% filenames) # read peter's data and keep keep subset of 766 we have thumbnails of

clf <- read.csv("../../API_tests/catal_sample_Clarifi_predictions.csv")

tdf <- clf %>% 
  rename(score = value,     # renaming 
         value = name,
         filename = file_name) %>%
  mutate(type = "CP_label",     # add type: CP_label
         score = round(score, 2)) %>% # reduce to 2 digits
  select(type, value, score, filename)

preds <- c_gviz_sub %>% 
  mutate(type = as.character(type), # renaming
         type = ifelse(type == "entity", "GV_entity", (ifelse(type == "label", "GV_label", "GV_ocr")))) %>% 
  select(type, value, score, filename) %>% 
  rbind(tdf) # combine CP and GV
```

# Images 101-200 

```{r, echo=FALSE, results='asis'}
# create a table with kable and this:
#https://stackoverflow.com/questions/25106481/add-an-image-to-a-table-like-output-in-r
# add image and format using kableExtra:
# https://haozhu233.github.io/kableExtra/awesome_table_in_html.html

for(f in filenames[101:200]){
  preds %>% 
    filter(filename == f) %>%   # select one image
    # add pointer to thumb
    mutate(img = sprintf('![](./thumbs/thumb_%s)', f)) %>% 
    select(-filename) %>%  # remove image name, we'll put it in the caption
    # print table
    kable(caption = f) %>%
      column_spec(4, width = "330px") %>% 
      collapse_rows(columns = 4, valign = "top") %>% 
    print() # need to explicitly print from for loop
}
```