假设我有一个由点组成的数据框:
DF1:
x y z label
1.1 2.1 3.1 2
4.1 5.1 6.1 1
7.1 8.1 9.1 0
我还有另一个点数据框:
DF2:
x y z label
4 5 6 0
7 8 9 1
1 2 3 2
无论如何都要通过df一个,看看哪个点最接近df2的内部,并将标签替换为它最接近的点的标签。
结果我想:
x y z label
1.1 2.1 3.1 2
4.1 5.1 6.1 0
7.1 8.1 9.1 1
谢谢你阅读我的问题!
我只能想到来自distance
的scipy
from scipy.spatial import distance
df1['label']=df2.label.iloc[distance.cdist(df1.iloc[:,:-1], df2.iloc[:,:-1], metric='euclidean').argmin(1)].values
df1
Out[446]:
x y z label
0 1.1 2.1 3.1 2
1 4.1 5.1 6.1 0
2 7.1 8.1 9.1 1
这是一个使用kd-tree的版本,对于大型数据集来说可能要快得多。
import numpy as np
import pandas as pd
from sklearn.neighbors import KDTree
np.random.seed(0)
#since you have df1 and df2, you will want to convert the dfs to array here with
#X=df1['x'.'y','z'].to_numpy()
#Y=df2['x','y','z'].to_numpy()
X = np.random.random((10, 3)) # 10 points in 3 dimensions
Y = np.random.random((10, 3))
tree = KDTree(Y, leaf_size=2)
#loop though the x array and find the closest point in y to each x
#note the you can find as many as k nearest neighbors by this method
#though yours only calls for the k=1 case
dist, ind = tree.query(X, k=1)
df1=pd.DataFrame(X, columns=['x','y','z'])
#set the labels to the closest point to each neighbor
df1['label']=ind
#this is cheesy, but it removes the list brackets
#get rid of the following line if you want more than k=1 nearest neighbors
df1['label']=df1['label'].str.get(0).str.get(0)
print(df1)
df1:
x y z
0 0.548814 0.715189 0.602763
1 0.544883 0.423655 0.645894
2 0.437587 0.891773 0.963663
3 0.383442 0.791725 0.528895
4 0.568045 0.925597 0.071036
5 0.087129 0.020218 0.832620
6 0.778157 0.870012 0.978618
7 0.799159 0.461479 0.780529
8 0.118274 0.639921 0.143353
9 0.944669 0.521848 0.414662
df2:
x y z
0 0.264556 0.774234 0.456150
1 0.568434 0.018790 0.617635
2 0.612096 0.616934 0.943748
3 0.681820 0.359508 0.437032
4 0.697631 0.060225 0.666767
5 0.670638 0.210383 0.128926
6 0.315428 0.363711 0.570197
7 0.438602 0.988374 0.102045
8 0.208877 0.161310 0.653108
9 0.253292 0.466311 0.244426
Out:
x y z label
0 0.548814 0.715189 0.602763 0
1 0.544883 0.423655 0.645894 6
2 0.437587 0.891773 0.963663 2
3 0.383442 0.791725 0.528895 0
4 0.568045 0.925597 0.071036 7
5 0.087129 0.020218 0.832620 8
6 0.778157 0.870012 0.978618 2
7 0.799159 0.461479 0.780529 2
8 0.118274 0.639921 0.143353 9
9 0.944669 0.521848 0.414662 3
这是使用matplotlib 3.0.2版的绘图代码
fig = plt.figure()
ax = Axes3D(fig)
ax.scatter(X[:,0],X[:,1],X[:,2])
ax.scatter(Y[:,0],Y[:,1],Y[:,2])
for i in range(len(X)): #plot each point + it's index as text above
ax.text(X[i,0],X[i,1],X[i,2], '%s' % (str(i)), size=20, zorder=1, color='blue')
for i in range(len(Y)): #plot each point + it's index as text above
ax.text(Y[i,0],Y[i,1],Y[i,2], '%s' % (str(i)), size=20, zorder=1, color='orange')
SELECT ABS($df1 - $df2) as nearest, ...
FROM yourtable
ORDER BY nearest ASC
LIMIT 1
通过'X'索引对它们进行排序,然后比较$ results数组,这将查找表之间最接近的数字。
https://www.w3schools.com/sql/func_sqlserver_abs.asp ABS函数返回一个绝对数字,因此只要你在df2上有整数就可以得到一个很好的解决方案。
希望能帮助到你。
我的第一个答案解决了所提出的问题,但OP想要一个针对任意数量维度的通用解决方案,而不仅仅是三个维度。
import numpy as np
import pandas as pd
from sklearn.neighbors import KDTree
np.random.seed(0)
#since you have df1 and df2, you will want to convert the dfs to array here with
#X=df1['x'.'y','z'].to_numpy()
#Y=df2['x','y','z'.to_numpy()
n=11 #n=number of dimensions in your sample
X = np.random.random((10, n)) # 10 points in n dimensions
Y = np.random.random((10, n))
tree = KDTree(Y, leaf_size=2)
indices=[]
#for i in range(len(X)):
#loop though the x array and find the closest point in y to each x
dist, ind = tree.query(X, k=1)
#indices.append(ind)
df1=pd.DataFrame(X)
##set the labels to the closest point to each neighbor
df1['label']=ind
你想要的结果现在是df1,但是你不能轻易地绘制它,或者在没有疯狂大脑的情况下解释它。基于3d版本的成功证明也在此处发布。