pandas 中的分类数据

在 Python 中处理分类数据

Kasey Jones

Research Data Scientist

dtypes：object

adult = pd.read_csv("data/adult.csv")
adult.dtypes

Age                 int64
Workclass          object
fnlgwt              int64
Education          object
Education Num       int64
Marital Status     object
Occupation         object
Relationship       object
...

dtypes：categorical

默认 dtype：

adult["Marital Status"].dtype

dtype('O')

设为分类：

adult["Marital Status"] = adult["Marital Status"].astype("category")

adult["Marital Status"].dtype

CategoricalDtype(categories=[' Divorced', ' Married-AF-spouse',
' Married-civ-spouse', ' Married-spouse-absent', ' Never-married',
' Separated', ' Widowed'], ordered=False)

创建分类 Series

my_data = ["A", "A", "C", "B", "C", "A"]

my_series1 = pd.Series(my_data, dtype="category")
print(my_series1)

0    A
1    A
2    C
...
dtype: category
Categories (3, object): [A, B, C]

创建分类 Series

my_data = ["A", "A", "C", "B", "C", "A"]
my_series2 = pd.Categorical(my_data, categories=["C", "B", "A"], ordered=True)
my_series2

[A, A, C, B, C, A]
Categories (3, object): [C < B < A]

为何用分类：内存

内存优化：

adult = pd.read_csv("data/adult.csv")
adult["Marital Status"].nbytes

adult["Marital Status"] = adult["Marital Status"].astype("category")
adult["Marital Status"].nbytes

读取数据时指定 dtypes

创建字典：

adult_dtypes = {"Marital Status": "category"}

设置 dtype 参数：

adult = pd.read_csv("data/adult.csv", dtype=adult_dtypes)

检查 dtype：
```
adult["Marital Status"].dtype
```

CategoricalDtype(categories=[' Divorced', ' Married-AF-spouse',
                             ..., ' Widowed'], ordered=False)

pandas 分类练习

在 Python 中处理分类数据