Google 使用 python 数据帧计算位置距离

Question

我正在尝试使用 Google 位置数据和 Python Pandas 数据框来梳理我在某个区域（一英里左右）的日期。

首先从latitudeE7转换为纬度：

with open(Takeout_google_location_history) as f:
data = json.loads(f.read())

df = json_normalize(data['locations'])
df['latitudeE7'] = df['latitudeE7'].div(10000000.0)
df['longitudeE7'] = df['longitudeE7'].div(10000000.0)
df.head()

然后计算距离：

import haversine as hs
from haversine import Unit
loc1 = (31.393300,-99.070050)
df['diff'] = hs.haversine(loc1,(df['latitudeE7'],df['longitudeE7']),unit=Unit.MILES)
df.head()

出现此错误：

~\Anaconda2\envs\notebook\lib\site-packages\haversine\haversine.py in 
haversine(point1, point2, unit)
     92     lat1 = radians(lat1)
     93     lng1 = radians(lng1)
---> 94     lat2 = radians(lat2)
     95     lng2 = radians(lng2)
     96 

~\Anaconda2\envs\notebook\lib\site-packages\pandas\core\series.py in             wrapper(self)
    183         if len(self) == 1:
    184             return converter(self.iloc[0])
--> 185         raise TypeError(f"cannot convert the series to {converter}")
    186 
    187     wrapper.__name__ = f"__{converter.__name__}__"

TypeError: cannot convert the series to <class 'float'>      

我不确定如何处理数据才能使其成为浮点数。

我试过：

df['latitudeE7'] = df['latitudeE7'].div(10000000.0).astype(float)

以及使用手写距离：

import math
def distance(origin, destination):

  lat1, lon1 = origin
  lat2, lon2 = destination
  radius = 6371  # km

  dlat = math.radians(float(lat2) - lat1)
  dlon = math.radians(float(lon2) - lon1)
  a = (math.sin(dlat / 2) * math.sin(dlat / 2) +
     math.cos(math.radians(lat1)) * math.cos(math.radians(lat2)) *
     math.sin(dlon / 2) * math.sin(dlon / 2))
  c = 2 * math.atan2(math.sqrt(a), math.sqrt(1 - a))
  d = radius * c

  return d

仍然出现同样的错误：

~\AppData\Local\Temp/ipykernel_22916/3664391511.py in distance(origin, destination)
     26     radius = 6371  # km
     27 
---> 28     dlat = math.radians(float(lat2) - lat1)
     29     dlon = math.radians(float(lon2) - lon1)
     30     a = (math.sin(dlat / 2) * math.sin(dlat / 2) +

~\Anaconda2\envs\notebook\lib\site-packages\pandas\core\series.py in wrapper(self)
    183         if len(self) == 1:
    184             return converter(self.iloc[0])
--> 185         raise TypeError(f"cannot convert the series to {converter}")
    186 
    187     wrapper.__name__ = f"__{converter.__name__}__"

TypeError: cannot convert the series to <class 'float'>

Answer 1

您不能直接将 pd.Series 传递给 haversine 函数。

代码：

from haversine import haversine, Unit
import pandas as pd

loc1 = (31.393300, -99.070050)

# Sample dataframe
df = pd.DataFrame({'latitudeE7': [0, 0], 'longitudeE7': [0, 0]})

# Calculation
# df['diff'] = haversine(loc1, (df['latitudeE7'], df['longitudeE7']), unit=Unit.MILES) # This doesn't work
df['diff'] = df.apply(lambda row: haversine(loc1, (row['latitudeE7'], row['longitudeE7']), unit=Unit.MILES), axis=1)

输出：

latitudeE7	longitudeE7	diff
0	0	6752.74
0	0	6752.74

参考：

您遇到的问题似乎与以下问题有关 post：understanding math errors in pandas dataframes

---

[编辑]

如果行数较多，haversin_vector速度较快

代码

#准备工作：

from haversine import haversine, haversine_vector, Unit
import pandas as pd
import numpy as np

loc1 = (31.393300, -99.070050)

# Sample dataframe
n = 1000000
df = pd.DataFrame({'latitudeE7': np.random.rand(n) * 180 - 90, 'longitudeE7': np.random.rand(n) * 360 - 180})

# 速度测试 1（使用 haversine）

df['diff'] = df.apply(lambda row: haversine(loc1, (row['latitudeE7'], row['longitudeE7']), unit=Unit.MILES), axis=1)

9.9 s ± 172 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

#速度测试2（使用haversine_vector）

df['diff'] = haversine_vector(loc1, df, unit=Unit.MILES, comb=True)

105 ms ± 1.1 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)

参考：

• haversine_vector: document
• haversine_vector: implementation