python - How to convert string to radians in pandas to calculate distance between two points

Question

I have a dataframe df:

{'city': {0: 'Adak', 1: 'Akiachak', 2: 'Akiak', 3: 'Akutan', 4: 'Alakanuk'},
 'latitudedegrees': {0: '51.87957',
  1: '60.88981',
  2: '60.911865',
  3: '54.098693',
  4: '62.683391'},
 'latituderadians': {0: 0.9054693110188746,
  1: 1.0627276654137685,
  2: 1.0631125977802958,
  3: 0.9442003138756087,
  4: 1.094031559264981},
 'longitudedegrees': {0: '-176.63675',
  1: '-161.42393',
  2: '-161.22577',
  3: '-165.88176',
  4: '-164.65455'},
 'longituderadians': {0: -3.082892867522094,
  1: -2.8173790700088506,
  2: -2.8139205255630984,
  3: -2.8951828810030293,
  4: -2.8737640258896295},
 'ncity': {0: 'Dallas', 1: 'Dallas', 2: 'Dallas', 3: 'Dallas', 4: 'Dallas'},
 'nlatituderadians': {0: 0.5722195078367402,
  1: 0.5722195078367402,
  2: 0.5722195078367402,
  3: 0.5722195078367402,
  4: 0.5722195078367402},
 'nlongituderadians': {0: -1.6891776914122487,
  1: -1.6891776914122487,
  2: -1.6891776914122487,
  3: -1.6891776914122487,
  4: -1.6891776914122487},
 'nstate': {0: 'TX', 1: 'TX', 2: 'TX', 3: 'TX', 4: 'TX'},
 'state': {0: 'AK', 1: 'AK', 2: 'AK', 3: 'AK', 4: 'AK'},
 'zip': {0: '99546', 1: '99551', 2: '99552', 3: '99553', 4: '99554'}}

it's a cartesian product of a list of 'ncity', and is several millions rows. The original file is here:

https://public.opendatasoft.com/explore/dataset/us-zip-code-latitude-and-longitude/export/

df already has radians, but they are being brought in as strings, and therefore do not run in here:

def distanceBetweenCityInMiles(lat1, long1, lat2, long2): # assumes latitudes and longitudes are in radians
    d = np.arccos(np.sin(lat1)*np.sin(lat2)+np.cos(lat1)*np.cos(lat2)*np.cos(long1-long2))
    distance_km = 6371 * d # distance_km ≈ radius_km * distance_radians ≈ 6371 * d, where 6371 km is the average radius of the earth
    distance_mi = distance_km * 0.621371
    return distance_mi

I've tried converting to float:

df[['nlatituderadians','nlongituderadians','latituderadians','longituderadians']]=df[['nlatituderadians','nlongituderadians','latituderadians','longituderadians']].astype(float)

But still get this error:

df['ncitydistance']= distanceBetweenCityInMiles('nlatituderadians', 'nlongituderadians', 'latituderadians', 'longituderadians')

TypeError: ufunc 'sin' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''

As you can see I have all the data in one row, and need to calculate the distance between the nlat/nlong and lat/long values.

How can I convert this string to radians to run the data through the distance function? I am assuming that is the reason this will not work. The end result should be another column that gives the distance between the cities.

Answer 1

You had some issues with your function. You only had the name of the column but did not specify a dataframe. For example, instead of just putting the column name variable lat1, you have to put the column name as enclosed brackets next to dataframe to actually call the column and not just a string with: df[lat1]:

def distanceBetweenCityInMiles(df, lat1, long1, lat2, long2): # assumes latitudes and longitudes are in radians
    d = np.arccos(np.sin(df[lat1])*np.sin(df[lat2])+np.cos(df[lat1])*np.cos(df[lat2])*np.cos(df[long1]-df[long2]))
    distance_km = 6371 * d # distance_km ≈ radius_km * distance_radians ≈ 6371 * d, where 6371 km is the average radius of the earth
    distance_mi = distance_km * 0.621371
    return distance_mi


df['ncitydistance'] =  distanceBetweenCityInMiles(df, 'nlatituderadians', 'nlongituderadians', 'latituderadians', 'longituderadians')
df['ncitydistance']
Out[1]: 
0    4065.460680
1    3426.266819
2    3419.729121
3    3598.672064
4    3538.417833
Name: ncitydistance, dtype: float64