Question

我有一个功能，它预先处理包含该信息的{csv}文件中的数据http://archive.ics.uci.edu/ml/machine-learning-databases/heart-disease/processed.cleveland.data

def loadAndPreprocess(filename):
    # ------------------------------------------------------------------------------------------------
    print("Activity 1a: ")

    dataLabels = [
    'age',
    'sex',
    'cp',
    'trestbps',
    'chol',
    'fbs',
    'restecg',
    'thalach',
    'exang',
    'oldpeak',
    'slope',
    'ca',
    'thal',
    'class'
    ]


    data=pd.read_csv(filename,sep=',',header=None,names=dataLabels,na_values=["?"])
    print(data)

    #print("Original data")
    #print(data.ix[:,0:])

    n=len(data)
    #print ("Amount of instances: " + str(n))
    #print ("Analyzing class distribution")
    #print (list(data['class'].value_counts()))

    #print("Rows with missing values")
    #print(sum(numpy.isnan(data).any(axis=1)))
    #print("Attributes with missing values")
    #print(len(data.isnull().sum().loc[data.isnull().sum()> 0]))


    #Remove rows with missing data
    cleanData=data[~numpy.isnan(data).any(axis=1)]
    cleanData=cleanData.reset_index(drop=True)  #Required. Otherwise, the index of the rows dropped keep active

    #print("Clean data")
    #print(cleanData[:])


    # Separating classes (Y) from values (X)
    dataX=cleanData.ix[:,0:13]
    dataY=cleanData.ix[:,13]


    # Extract status and standardize product values


    attributes = preprocessing.scale(dataX)
    #print("Scaled data")
    #print(attributes[:])
    return attributes, dataY

def exercise1(attributes, classes):
    print("Activity 1a")

    # Apply PCA requesting all components (no argument)
    pca = PCA(n_components=3)
    pca.fit(attributes)

    result = pca.transform(attributes)

    fig = plt.figure(1, figsize=(10, 6))
    sp = fig.gca(projection='3d')
    sp.scatter(attributes[:, 0], attributes[:, 1], attributes[:, 2])
    plt.show()

    X, y = loadAndPreprocess('processedCleveland.csv')
    exercise1(X, y)

但是当我显示结果时，会分为两组，一组靠近y轴的最大值，另一组靠近y轴的最小值，形成水平线。

我认为我没有以正确的方式传递数据，但我不确定是什么问题。

Answer 1

请更改以下行

sp.scatter(attributes[:, 0], attributes[:, 1], attributes[:, 2])

到

sp.scatter(result[:, 0], result[:, 1], result[:, 2])

使用来自sklearn的PCA的意外结果

1 个答案: