Das deutsche Python-Forum

Code: Alles auswählen

#!/usr/bin/env python
FILE_CORRECT = 'correct.txt'
FILE_FAILURE = 'failure.txt'

from collections import defaultdict

import nltk


def ngrams(text, n):
    liste = []
    if n < len(text):
        for p in range(len(text) - n + 1) :
            tg = text[p:p+n]
            liste.append(tg)
    return liste


def xgrams(text, maxlen=3):
    return [w for n in range(1, maxlen+1) for w in ngrams(text.lower(), n)]


def extract_features(text):
    rv = defaultdict(lambda: 0)
    for w in xgrams(text.lower()): rv[w] += 1
    return dict(rv)
    

if __name__ == '__main__':
    with open(FILE_CORRECT, 'r') as ifh: data_correct = ifh.read().splitlines()
    with open(FILE_FAILURE, 'r') as ifh: data_failure = ifh.read().splitlines()
    
    classifier = nltk.NaiveBayesClassifier.train([(extract_features(line), line) for line in data_correct])
    for line in data_failure:
        classified = classifier.classify(extract_features(line))
        probdist = classifier.prob_classify(extract_features(line))
        print('{:10s} | {:0.2f} | {}'.format(classified, probdist.prob(classified), line))

Code: Alles auswählen

Nike       | 1.00 | Nik e
Nike       | 1.00 | Nik
Bosch AG   | 1.00 | bosch ag
Bosch AG   | 1.00 | bosch ag gmbh
Bosch AG   | 1.00 | B0sch AG

Code: Alles auswählen

#!/usr/bin/env python
from collections import defaultdict

import nltk
from nltk.util import ngrams


FILE_CORRECT = "correct.txt"
FILE_FAILURE = "failure.txt"


def extract_features(text):
    features = defaultdict(int)
    for count in range(1, 4):
        for ngram in ngrams(text, count):
            features[ngram] += 1
    return features


if __name__ == "__main__":
    with open(FILE_CORRECT, "r") as text_file:
        data_correct = text_file.read().splitlines()
    with open(FILE_FAILURE, "r") as text_file:
        data_failure = text_file.read().splitlines()

    classifier = nltk.NaiveBayesClassifier.train(
        ((extract_features(line.lower()), line) for line in data_correct)
    )
    for line in data_failure:
        classified = classifier.classify(extract_features(line))
        probdist = classifier.prob_classify(extract_features(line))
        print(f"{classified:10s} | {probdist.prob(classified):0.2f} | {line}")

Code: Alles auswählen

pd.read_excel()

Code: Alles auswählen

FILE_CORRECT =df1['Name'] 
FILE_FAILURE= df2['Name'] 

Code: Alles auswählen

if __name__ == "__main__":
    with open(FILE_CORRECT, "r") as text_file:
        data_correct = text_file.read().splitlines()
    with open(FILE_FAILURE, "r") as text_file:
        data_failure = text_file.read().splitlines()

Code: Alles auswählen

x = df_correct['name'].to_string().splitlines()

Code: Alles auswählen

def predict_name(val): 
    # Korrekten Namen 'name' und Wahrscheinlichkeit 'prob' berechnen
    return pd.Series({'name': name, 'prob': prob})

df[['name', 'prob']] = df.Marken.apply(predict_name, axis=1)