熊猫计算正/负/中性值

Question

In Python Pandas, I have a data frame with columns and records in the following format:

text           source    senti
-------------------------------
great food     site1     0.6
awful staff    site4     -0.4
good chef      site8     0.4
average food   site6     0.05
bad food       site2     -0.8

The text column is essentially a description or opinion of something. I want to draw some conclusions about average sentiment on the sets of data, with the output like this.

sentiment    count
----------------
positive     2
neutral      1
negative     2

Where we have a count of 'senti' grouped as positive, negative or neutral.

The sentiments are counted as each group upon meeting the following conditions:

• A positive record has sentiment >0.1
• Neutral records have scores >-0.1 AND <0.1
• Negative records have scores <-0.1

Big thanks in advance

Answer 1

I'd use pd.cut + groupby

cut = pd.cut(
    df.senti,
    [-np.inf, -.1, .1, np.inf],
    labels=['positive', 'neutral', 'negative']
)

df.groupby(cut).senti.count().reset_index(name='count')

      senti  count
0  positive      2
1   neutral      1
2  negative      2

---

As pointed out by @root, pd.value_counts gives the same solution on the cut variable.

pd.value_counts(cut, sort=False).rename_axis('senti').reset_index(name='count')

Answer 2

使用的另一个版本apply于映射到组：

df.groupby(df['senti'].apply(lambda x: 'negative' if x < -0.1 else 'positive' if x > 0.1 else 'neutral'))['senti'].count()