python中KNN的数据预处理

Question

预处理要花很多时间去理解，tuple，list，float，array结构。 我的数据看起来像

<bound method NDFrame.head of                                                       X                                 Y
0     [1.9902, 1.9902, 1.9902, 1.9902, 1.9902, 0.034...      [0.097, 0.097, 0.097, 0.094]
1     [1.9902, 0.034, 0.034, 0.034, 0.034, 0.034, 0....      [0.094, 0.094, 0.094, 0.094]
2     [0.034, 0.034, 0.097, 0.097, 0.097, 0.097, 0.0...  [1.0882, 1.0882, 1.0882, 1.0882]
3     [0.097, 0.097, 0.097, 0.094, 0.094, 0.094, 0.0...  [1.0882, 1.2382, 1.2382, 1.2382]
4     [0.094, 0.094, 0.094, 0.094, 1.0882, 1.0882, 1...  [1.2382, 1.2382, 1.2182, 1.2182]
...                                                 ...                               ...
3395  [0.136, 0.286, 0.286, 0.286, 0.286, 0.286, 0.2...  [0.1276, 0.1276, 0.1276, 0.1276]
3396  [0.286, 0.286, 0.266, 0.266, 0.266, 0.266, 0.2...   [1.1423, 1.2923, 1.2723, 3.672]
3397  [0.266, 0.266, 0.266, 0.1276, 0.1276, 0.1276, ...      [3.672, 3.672, 3.772, 3.772]
3398  [0.1276, 0.1276, 0.1276, 0.1276, 1.1423, 1.292...      [3.772, 3.802, 3.802, 3.802]
3399  [1.1423, 1.2923, 1.2723, 3.672, 3.672, 3.672, ...      [1.021, 1.021, 1.021, 1.021]

我正在使用

进行数据拆分

x=csv_data['X']
y=csv_data['Y']
#print(x)
x_train, x_test, y_train, y_test = train_test_split(x,y)

拟合 KNN 模型

K = []
training = []
test = []
scores = {}
  
for k in range(2, 21):
    clf = KNeighborsClassifier(n_neighbors = k)
    clf.fit(x_train, y_train)
  
    training_score = clf.score(x_train, y_train)
    test_score = clf.score(x_test, y_test)
    K.append(k)
  
    training.append(training_score)
    test.append(test_score)
    scores[k] = [training_score, test_score]

遇到错误

TypeError                                 Traceback (most recent call last)
TypeError: float() argument must be a string or a number, not 'list'

The above exception was the direct cause of the following exception:

ValueError                                Traceback (most recent call last)
<ipython-input-93-906aa771beda> in <module>()
      6 for k in range(2, 21):
      7     clf = KNeighborsClassifier(n_neighbors = k)
----> 8     clf.fit(x_train, y_train)
      9 
     10     training_score = clf.score(x_train, y_train)

7 frames
/usr/local/lib/python3.7/dist-packages/numpy/core/_asarray.py in asarray(a, dtype, order)
     81 
     82     """
---> 83     return array(a, dtype, copy=False, order=order)
     84 
     85 

ValueError: setting an array element with a sequence.

我一直在尝试一些方法，例如 preprocessing 或 StandardScaler dint 对我有用。 请帮助运行 KNN。 谢谢

Answer 1

问题是在使用KNN 时，您的y 的形状为(n, 4)，而KNN.fit 方法希望您的y 的形状为(n,1)。所以简而言之，您只能从y 中预测 1 个值。所以简而言之，对于y 中的每一列，您要么使用KNN 4 次，要么不使用KNN。

代码会是这样的

# Import KNN for regression

y1 = y.iloc[:, 0]
y2 = y.iloc[:, 1]
y3 = y.iloc[:, 2]
y4 = y.iloc[:, 3]

regressor1 = KNeighborsRegressor(n_neighbors=k).fit(x, y1)
regressor2 = KNeighborsRegressor(n_neighbors=k).fit(x, y2)
regressor3 = KNeighborsRegressor(n_neighbors=k).fit(x, y3)
regressor4 = KNeighborsRegressor(n_neighbors=k).fit(x, y4)

天啊！！现在我看到您使用KNN 进行分类，实际上您的问题是回归。你的基础真的很差。

另外，甚至不要使用它。你不会从中得到任何好的结果，而且计算成本也很高。