Loading a Dataset for Linear SVM Classification from a CSV file

问题

I have a csv file below called train.csv:

   25.3, 12.4, 2.35, 4.89, 1, 2.35, 5.65, 7, 6.24, 5.52, M
   20, 15.34, 8.55, 12.43, 23.5, 3, 7.6, 8.11, 4.23, 9.56, B
   4.5, 2.5, 2, 5, 10, 15, 20.25, 43, 9.55, 10.34, B
   1.5, 2.5, 3.5, 4.5, 5.5, 6.5, 7.5, 8.5, 9.5, 10.5, M

I am trying to get this dataset be separated and classified as the following (This is the output I want):

    [[25.3, 12.4, 2.35, 4.89. 1, 2.35, 5.65, 7, 6.24, 5.52], 
    [20, 15.34, 8.55, 12.43, 23.5, 3, 7.6, 8.11, 4.23, 9.56], 
    [4.5, 2.5, 2, 5, 10, 15, 20.25, 43, 9.55, 10.34], 
    [1.5, 2.5, 3.5, 4.5, 5.5, 6.5, 7.5, 8.5, 9.5, 10.5]], 
    [M, B, B, M]

The one in "[[" is the x (the sample data) and the one in "[M, M, B, B, M]" is the y (which is the classification that matches with its set of data.

I am trying to create a python code that's been loaded and can print out the data being separated by data and it's classification. It's related to linear SVM.

y_list = []
x_list = []
for W in range(0, 100):
    X = data_train.readline()
    y = X.split(",")
    y_list.append(y[10][0])
    print(y_list)
    z_list = []
    for Z in range(0, 10):
        z_list.append(y[Z])
    x_list.append(z_list)
    dataSet = (x_list, y_list)
    print(dataSet)

Note: I know my range is completely wrong. I'm not sure how to fit the range at all for this type of example, could anyone please explain how the range would work in this situation.

Note: I know the append line where it is "y[10][0]" is also wrong as well. Could someone explain how these indexes work.

Overall I want the output to be the output I stated above. Thanks for the help.

回答1:

First, I think you have an error in your CSV in the first row:

25.3, 12.4, 2.35, 4.89. 1, 2.35, 5.65, 7, 6.24, 5.52, M

I just assumed it should be 4.89, 1, and not 4.89. 1.

Second, I recommend you to use pandas to read that CSV, and then do this:

import pandas as pd
data = pd.read_csv('prueba.csv', header=None, usecols=[i for i in range(11)])
# the usecols=[i for i in range(11)] will create a list of numbers for your columns
# that line will make a dataframe called data, which will contain your data.
l = [i for i in range(10)]
X_train = data[l]
y_train = data[10]

This is the most easy way to have ready your data for any machine learning algorithm in scikit-learn.

回答2:

import pandas as pd

df = pd.read_csv(/path/to/csv, header=None, index_col=False)
x = df.iloc[:,:-1].values
y = df.iloc[:,-1:].values

回答3:

I think you should use pandas, which is a library that helps you with reading csv:

import pandas as pd

dataset = pd.read_csv('train.cvs')

Second you can use train_test_split to automatically split the data:

X_train, X_test, y_train, y_test = train_test_split(
        X, y, stratify=y, test_size=0.2)

This will split the data where X_train and X_test comprises of 80% of the data and y_train, y_test 20%. This can be changed with adjusting test_size. stratify will automatically make the ratio of classification count (M, B) equal in train and test, which is generally considered good practise in machine learning. This will generate random split each time. If you want the same split you can use random_state=(SEED) as keyword argument.

After that you can continue on with the machine learning:

from sklearn.svm import SVC
from sklearn.metrics import confusion_matrix, classification_report

# Important to scale
sc = StandardScaler()
X_train = sc.fit_transform(X_train)
X_test = sc.transform(X_test)

clf = SVC()
clf.fit(X_train, y_train)
pred = clf.predict(X_test)

print(classification_report(y_test, pred))
print(confusion_matrix(y_test, pred))

来源：https://stackoverflow.com/questions/58963975/loading-a-dataset-for-linear-svm-classification-from-a-csv-file

标签

python

svm

libsvm