在 Pandas Dataframe 中查找最接近输入的值

Question

The first rows of my dataframe are shown below. The columns are longitude, latitude, and value. This dataframe extends for 30 million rows.

-179.979166666666657 89.9791666666666714 -3.39999995214436425e+38
-179.9375 89.9791666666666714 -3.39999995214436425e+38
-179.895833333333343 89.9791666666666714 -3.39999995214436425e+38
-179.854166666666657 89.9791666666666714 -3.39999995214436425e+38
-179.8125 89.9791666666666714 -3.39999995214436425e+38
-179.770833333333343 89.9791666666666714 -3.39999995214436425e+38
-179.729166666666657 89.9791666666666714 -3.39999995214436425e+38
-179.6875 89.9791666666666714 -3.39999995214436425e+38
-179.645833333333343 89.9791666666666714 -3.39999995214436425e+38

I am trying to find the closest longitude and latitude point to a given input, and then print out the value associated with the closest longitude and latitude. I have tried to convert the dataframe into an array, and then search for the minimum value using this algorithm:

def match (lon, lat):
    min=10000
    minindex=-1
    for x in range (len (mintemparr)):
        if (abs ((float (lon))-float (mintemparr [x][0])))+(abs ((float (lat))-float (mintemparr [x] 
        [1])))<min:
             
              min=(abs ((float (lon))-float (mintemparr [x][0])))+(abs ((float (lat))-float 
             (mintemparr [x][1])))
       minindex=x
    result=mintemparr [minindex][2]
    return result 

However, this is very slow. Is there a more direct way to search for the closest value within pandas rather than converting it into an array.

Thanks in advance.

Answer 1

def find_closest(df, lat, lon):
    dist = (df['lat'] - lat).abs() + (df['lon'] - lon).abs()
    return df.loc[dist.idxmin()]

>>> find_closest(df, -179, 90)
lat     -1.796458e+02
lon      8.997917e+01
value   -3.400000e+38
Name: 8, dtype: float64

Answer 2

You can do this by using pandas. I will make a column for the differences with the given values and then the square root of the sum of squared differences of latitude and longitude. Then I'm getting the min. Assuming that your data frame is called df with columns latitude, longitude and value :

lon=-179.979164

lat=89.979162

df['sumofdiff']=df.assign(landif=df['longitude']-lon).assign(latdiff=df['latitude']-lat).eval("x=(landif*landif)+(latdiff*latdiff)")['x'].apply(np.sqrt)

df[df.sumofdiff == df.sumofdiff.min()]


    longitude   latitude         value      diff  sumofdiff
0 -179.979167  89.979167 -3.400000e+38  0.000005   0.000005