I am trying to use an SVM model for machine learning, but it doesn't complete

Question

I am trying to use an SVM model on a dataset. It has 125973 records in the train and 22544 records in the test. It keeps computing forever and never completes. Can anyone help me. Below is my python code so far..

import pandas as pd
import sklearn.svm as s
import sklearn.metrics as m
import sklearn.preprocessing as pp

traindata = pd.read_csv("E:\\a-train.csv")

testdata = pd.read_csv("E:\\a-test.csv")

for col in traindata.columns:
    if traindata[col].dtype == type(object):
        le = pp.LabelEncoder()
        traindata[col] = le.fit_transform(traindata[col])
        
for col in testdata.columns:
    if testdata[col].dtype == type(object):
        le = pp.LabelEncoder()
        testdata[col] = le.fit_transform(testdata[col])

countS = 0

featsTrain = traindata.values[:,0:13]

featsTest = testdata.values[:,0:13]

lblsTrain = traindata.values[:,13]

lblsTest = testdata.values[:,13]

modelS = s.SVC(cache_size = 7000)

modelS.fit(featsTrain, lblsTrain)

lblsPredS = modelS.predict(featsTest)

for a,b in zip(lblsTest, lblsPredS):
   if a == b:
      countS += 1

accS = (round(countS/(len(featsTest)), 3)) * 100

print( m.confusion_matrix(lblsTest, lblsPredS) )

print(m.classification_report(lblsTest, lblsPredS))

print("\nAccuracy = ", accS, "%")

what is the configuration of the server that you are using? this post might be helpful https://stackoverflow.com/questions/53940258/svc-classifier-taking-too-much-time-for-training/54004026 — Piyush Sambhi, Oct 19 '20 at 12:32

score 0 · Answer 1 · answered Oct 19 '20 at 12:35

0

Please go through this post, this might solve your issue/use case:

SVC classifier taking too much time for training

answered Oct 19 '20 at 12:35

Piyush Sambhi

843
4
13

I am trying to use an SVM model for machine learning, but it doesn't complete

1 Answers1