how to count specific words from a pandas Series?

问题

I am trying to count the number of keywords from a pandas DataFrame as such:

df = pd.read_csv('amazon_baby.csv')
selected_words = ['awesome', 'great', 'fantastic', 'amazing', 'love', 'horrible', 'bad', 'terrible', 'awful', 'wow', 'hate']

The selected_words have to be counted from the Series: df['review']

i have tried

def word_counter(sent):
a={}
for word in selected_words:
    a[word] = sent.count(word)
return a

and then

df['totalwords'] = df.review.str.split()
df['word_count'] = df.totalwords.apply(word_counter)

----------------------------------------------------------------------------
----> 1 df['word_count'] = df.totalwords.apply(word_counter)

c:\users\admin\appdata\local\programs\python\python36\lib\site-packages\pandas\core\series.py in apply(self, func, convert_dtype, args, **kwds)
   3192             else:
   3193                 values = self.astype(object).values
-> 3194                 mapped = lib.map_infer(values, f, convert=convert_dtype)
   3195 
   3196         if len(mapped) and isinstance(mapped[0], Series):

pandas/_libs/src\inference.pyx in pandas._libs.lib.map_infer()

<ipython-input-51-cd11c5eb1f40> in word_counter(sent)
  2     a={}
  3     for word in selected_words:
----> 4         a[word] = sent.count(word)
  5     return a

AttributeError: 'float' object has no attribute 'count'

can someone help..? i am guessing it is because of some fault value in the series that is not a string. . .

some people have tried helping but the issue is that the individual cells in the DataFrame have sentences in them.

I need to extract a count of selected words, preferably in dictionary form and store them in a new column in the same dataFrame with the corresponding rows.

Data in csv format

回答1:

Suppose your dataframe looks like this,

df=pd.DataFrame({'A': ['awesome', 'great', 'fantastic', 'amazing', 'love', 'horrible', 'bad', 'terrible', 'awful', 'wow', 'hate','great', 'fantastic', 'amazing', 'love', 'horrible']})
print(df)
    A
0   awesome
1   great
2   fantastic
3   amazing
4   love
5   horrible
6   bad
7   terrible
8   awful
9   wow
10  hate
11  great
12  fantastic
13  amazing
14  love
15  horrible

selected_words=['awesome','great','fantastic']

df.loc[df['A'].isin(selected_words),'A'].value_counts()
[out]
great        2
fantastic    2
awesome      1
Name: A, dtype: int64

回答2:

Repeated list.count in a loop would work, albeit inefficiently, with a list of values. Complexity would be O(m x n), where m is the number of selected values and n is the total number of values.

With Pandas you can use optimized methods which ensure O(n) complexity. In this case, you can use value_counts followed by reindex:

res = df['A'].value_counts().reindex(selected_words)

print(res)

awesome      1
great        2
fantastic    2
Name: A, dtype: int64

Or, as per @pyd's solution, filter first and then use value_counts. Both solutions will have O(n) complexity.

回答3:

In your question you seems to be implementing a dict for the count. @pyd have posted a good solution for the counting. The result produced is not a dict. If you are looking for a dictionary as output take a look at this code posted below, which is basically an extension of the solution provided by pyd.

df=pd.DataFrame({'A': ['awesome', 'great', 'fantastic', 'amazing', 'love', 'horrible', 'bad', 'terrible', 'awful', 'wow', 'hate','great', 'fantastic', 'amazing', 'love', 'horrible']})

def get_count_dict(data, selected_words):

    count_dict = {}

    counts = data.loc[data['A'].isin(selected_words), 'A'].value_counts()

    for i in range(len(counts.index.tolist())):
        count_dict[counts.index.tolist()[i]] = counts[i]

    return count_dict

selected_words=['awesome','great','fantastic']

get_count_dict(df, selected_words)

Output : {'fantastic': 2, 'great': 2, 'awesome': 1}

来源：https://stackoverflow.com/questions/52221931/how-to-count-specific-words-from-a-pandas-series

标签

python

python-3.x

pandas

attributeerror