如何将复杂的JSON数据转换为单个数据帧？

Question

我不知道如何正确地将我的JSON数据转换为有用的数据帧.这是显示我的数据结构的一些示例数据:

{
"data":[
{"track":[
{"time":"2015","midpoint":{"x":6,"y":8},"realworld":{"x":1,"y":3},"coordinate":{"x":16,"y":38}},
{"time":"2015","midpoint":{"x":6,"y":8},"realworld":{"x":1,"y":3},"coordinate":{"x":16,"y":37}},
{"time":"2016","midpoint":{"x":6,"y":9},"realworld":{"x":2,"y":3},"coordinate":{"x":16,"y":38}}
]},
{"track":[
{"time":"2015","midpoint":{"x":5,"y":9},"realworld":{"x":-1,"y":3},"coordinate":{"x":16,"y":38}},
{"time":"2015","midpoint":{"x":5,"y":9},"realworld":{"x":-1,"y":3},"coordinate":{"x":16,"y":38}},
{"time":"2016","midpoint":{"x":5,"y":9},"realworld":{"x":-1,"y":3},"coordinate":{"x":16,"y":38}},
{"time":"2015","midpoint":{"x":3,"y":15},"realworld":{"x":-9,"y":2},"coordinate":{"x":17,"y":38}}
]},
{"track":[
{"time":"2015","midpoint":{"x":6,"y":7},"realworld":{"x":-2,"y":3},"coordinate":{"x":16,"y":39}}
]}]}

我有很多曲目,我希望数据集看起来像这样:

track   time   midpoint   realworld   coordinate 
1
1
1
2
2
2
2
3

到目前为止我有这个:

json_file <- "testdata.json"
data <- fromJSON(json_file)
data2 <- list.stack(data, fill=TRUE) 

现在它出来像这样:

我怎样才能以正确的格式获得这个？

Answer 1

flatten = TRUE读取时添加参数fromJSON。这将为您提供一个嵌套列表，其中最深层次是三个数据框的列表。使用：

library(jsonlite)
# read the json
jsondata <- fromJSON(txt, flatten = TRUE)

# bind the dataframes in the nested 'track' list together    
dat <- do.call(rbind, jsondata$data$track)

# add a track variable
dat$track <- rep(1:length(jsondata$data$track), sapply(jsondata$data$track, nrow))

给出：

> dat
  time midpoint.x midpoint.y realworld.x realworld.y coordinate.x coordinate.y track
1 2015          6          8           1           3           16           38     1
2 2015          6          8           1           3           16           37     1
3 2016          6          9           2           3           16           38     1
4 2015          5          9          -1           3           16           38     2
5 2015          5          9          -1           3           16           38     2
6 2016          5          9          -1           3           16           38     2
7 2015          3         15          -9           2           17           38     2
8 2015          6          7          -2           3           16           39     3

另一种更短的方法是jsonlite结合使用rbindlistfromdata.table包：

library(jsonlite)
library(data.table)
# read the json
jsondata <- fromJSON(txt, flatten = TRUE)
# bind the dataframes in the nested 'track' list together 
# and include an id-column at the same time   
dat <- rbindlist(jsondata$data$track, idcol = 'track')

或以类似的方式bind_rows从dplyr包中：

library(dplyr)
dat <- bind_rows(jsondata$data$track, .id = 'track')

使用数据：

txt <- '{
"data":[
{"track":[
{"time":"2015","midpoint":{"x":6,"y":8},"realworld":{"x":1,"y":3},"coordinate":{"x":16,"y":38}},
{"time":"2015","midpoint":{"x":6,"y":8},"realworld":{"x":1,"y":3},"coordinate":{"x":16,"y":37}},
{"time":"2016","midpoint":{"x":6,"y":9},"realworld":{"x":2,"y":3},"coordinate":{"x":16,"y":38}}
]},
{"track":[
{"time":"2015","midpoint":{"x":5,"y":9},"realworld":{"x":-1,"y":3},"coordinate":{"x":16,"y":38}},
{"time":"2015","midpoint":{"x":5,"y":9},"realworld":{"x":-1,"y":3},"coordinate":{"x":16,"y":38}},
{"time":"2016","midpoint":{"x":5,"y":9},"realworld":{"x":-1,"y":3},"coordinate":{"x":16,"y":38}},
{"time":"2015","midpoint":{"x":3,"y":15},"realworld":{"x":-9,"y":2},"coordinate":{"x":17,"y":38}}
]},
{"track":[
{"time":"2015","midpoint":{"x":6,"y":7},"realworld":{"x":-2,"y":3},"coordinate":{"x":16,"y":39}}
]}]}'