将嵌套的 JSON 解析为数据框

Question

我正在尝试将 JSON 字符串解析为熊猫数据帧的最低粒度。

尝试

首先我尝试了 read_json：

jsonData = pd.read_json(apiRequest)

但是很大一部分数据仍然嵌套在 networkRank 下。

然后我尝试了json_normalize，但是这次我错过了更高一级的数据，例如纬度和经度。

result = json_normalize(json_data['networkRank'])

我也尝试解析“成”嵌套结构并从头开始构造数据框，但此代码导致错误：

result_nested = json_normalize(json_data, 'networkRank', ['longitude', 'latitude', ['networkRank', 'type3G', 'downloadSpeed']])

目标

将JSON数据解析成一个包含所有字段的平面表，即在图2的每一行数据中附加纬度、经度和距离数据。

JSON 字符串

{'apiVersion': '2',
 'distance': 10,
 'latitude': '-6.162959',
 'longitude': '35.751607',
 'networkRank': [{'networkId': '6402',
   'networkName': 'Vodacom',
   'type3G': {'averageRssiAsu': '9.5429091136',
    'averageRssiDb': '-69.5664329624972',
    'downloadSpeed': '1508.1304',
    'networkId': '6402',
    'networkName': 'Vodacom',
    'networkType': '3',
    'pingTime': '320.9600',
    'reliability': '0.804236452826138',
    'sampleSizeRSSI': '948',
    'sampleSizeSpeed': '29',
    'uploadSpeed': '893.7692'}},
  {'networkId': '6400',
   'networkName': 'tiGO',
   'type3G': {'averageRssiAsu': '15.3537142857',
    'averageRssiDb': '-61.4563389583101',
    'downloadSpeed': '516.0000',
    'networkId': '6400',
    'networkName': 'tiGO',
    'networkType': '3',
    'pingTime': '259.0000',
    'reliability': '0.911904765537807',
    'sampleSizeRSSI': '935',
    'sampleSizeSpeed': '21',
    'uploadSpeed': '320.4211'}},
  {'networkId': '6403',
   'networkName': 'Airtel',
   'type3G': {'averageRssiAsu': '13.2729999375',
    'averageRssiDb': '-58.1521092977699',
    'downloadSpeed': '1080.2500',
    'networkId': '6403',
    'networkName': 'Airtel',
    'networkType': '3',
    'pingTime': '194.5556',
    'reliability': '0.554680264185345',
    'sampleSizeRSSI': '587',
    'sampleSizeSpeed': '21',
    'uploadSpeed': '572.1579'}}],
 'network_type': None,
 'perMinuteCurrent': 0,
 'perMinuteLimit': 10,
 'perMonthCurrent': 0,
 'perMonthLimit': 2000}

Answer 1

此函数递归调用自身以展平字典和列表。

from collections import OrderedDict

def flatten(json_object, container=None, name=''):
    if container is None:
        container = OrderedDict()
    if isinstance(json_object, dict):
        for key in json_object:
            flatten(json_object[key], container=container, name=name + key + '_')
    elif isinstance(json_object, list):
        for n, item in enumerate(json_object, 1):
            flatten(item, container=container, name=name + str(n) + '_')
    else:
        container[str(name[:-1])] = str(json_object)
    return container

例子：

flatten([1, 2, 3])
OrderedDict([('1', '1'), ('2', '2'), ('3', '3')])

flatten([1, 2, 3], name='x')
OrderedDict([('x1', '1'), ('x2', '2'), ('x3', '3')])

flatten({'a': [1, 2, 3], 'b': 4, 'c': {'d': [5, 6], 'e': 7}}, name='x')
OrderedDict([('xa_1', '1'),
             ('xa_2', '2'),
             ('xa_3', '3'),
             ('xc_e', '7'),
             ('xc_d_1', '5'),
             ('xc_d_2', '6'),
             ('xb', '4')])

回复：

# j = json string
>>> pd.DataFrame(flatten(j), index=[0]).T
                                                      0
perMinuteLimit                                       10
distance                                             10
perMonthCurrent                                       0
longitude                                     35.751607
perMonthLimit                                      2000
latitude                                      -6.162959
perMinuteCurrent                                      0
networkRank_1_networkId                            6402
networkRank_1_type3G_sampleSizeSpeed                 29
networkRank_1_type3G_averageRssiAsu        9.5429091136
networkRank_1_type3G_pingTime                  320.9600
networkRank_1_type3G_networkType                      3
networkRank_1_type3G_averageRssiDb    -69.5664329624972
networkRank_1_type3G_networkName                Vodacom
networkRank_1_type3G_networkId                     6402
networkRank_1_type3G_downloadSpeed            1508.1304
networkRank_1_type3G_uploadSpeed               893.7692
networkRank_1_type3G_reliability      0.804236452826138
networkRank_1_type3G_sampleSizeRSSI                 948
networkRank_1_networkName                       Vodacom
networkRank_2_networkId                            6400
networkRank_2_type3G_sampleSizeSpeed                 21
networkRank_2_type3G_averageRssiAsu       15.3537142857
networkRank_2_type3G_pingTime                  259.0000
networkRank_2_type3G_networkType                      3
networkRank_2_type3G_averageRssiDb    -61.4563389583101
networkRank_2_type3G_networkName                   tiGO
networkRank_2_type3G_networkId                     6400
networkRank_2_type3G_downloadSpeed             516.0000
networkRank_2_type3G_uploadSpeed               320.4211
networkRank_2_type3G_reliability      0.911904765537807
networkRank_2_type3G_sampleSizeRSSI                 935
networkRank_2_networkName                          tiGO
networkRank_3_networkId                            6403
networkRank_3_type3G_sampleSizeSpeed                 21
networkRank_3_type3G_averageRssiAsu       13.2729999375
networkRank_3_type3G_pingTime                  194.5556
networkRank_3_type3G_networkType                      3
networkRank_3_type3G_averageRssiDb    -58.1521092977699
networkRank_3_type3G_networkName                 Airtel
networkRank_3_type3G_networkId                     6403
networkRank_3_type3G_downloadSpeed            1080.2500
networkRank_3_type3G_uploadSpeed               572.1579
networkRank_3_type3G_reliability      0.554680264185345
networkRank_3_type3G_sampleSizeRSSI                 587
networkRank_3_networkName                        Airtel
network_type                                       None
apiVersion                                            2

Answer 2

1) 将 JSON 字符串解析为 python 结构

2) 遍历 'networkRank' 字典列表并将要添加的每个键放入哈希中

for data_row in deserialized_json['networkRank']:
    data_row['latitude'] = deserialized_json['latitude']
    # etc

3)

yourdataframe = pd.DataFrame( deserialized_json['networkRank'] )

Answer 3

这就是你想要的吗？

In [22]: df = json_normalize(json_data['networkRank'])

In [23]: df['distance'] = json_data['distance']

In [24]: df['latitude'] = json_data['latitude']

In [25]: df['longitude'] = json_data['longitude']

In [26]: df
Out[26]:
  networkId networkName type3G.averageRssiAsu type3G.averageRssiDb  \
0      6402     Vodacom          9.5429091136    -69.5664329624972
1      6400        tiGO         15.3537142857    -61.4563389583101
2      6403      Airtel         13.2729999375    -58.1521092977699

  type3G.downloadSpeed type3G.networkId type3G.networkName type3G.networkType  \
0            1508.1304             6402            Vodacom                  3
1             516.0000             6400               tiGO                  3
2            1080.2500             6403             Airtel                  3

  type3G.pingTime type3G.reliability type3G.sampleSizeRSSI  \
0        320.9600  0.804236452826138                   948
1        259.0000  0.911904765537807                   935
2        194.5556  0.554680264185345                   587

  type3G.sampleSizeSpeed type3G.uploadSpeed  distance   latitude  longitude
0                     29           893.7692        10  -6.162959  35.751607
1                     21           320.4211        10  -6.162959  35.751607
2                     21           572.1579        10  -6.162959  35.751607