作為在 Python 3.7 中運行 function 的一部分，我如何進行時間序列分析？

Question

我有這個：

formulas_count_stats.py：

import pandas as pd
from df_count_stats import df, df1

df = df
df1 = df1

class Data_load_compare_0:
    def __init__(self, df):
        self.df = pd.read_csv(df, delimiter=';')

        '''
        Data information section from df  = basic stats

        '''

    def get_EDA_columns(self):
        return self.df.columns

    def get_EDA_info(self):
        return self.df.info()

    def get_EDA_describe(self):
        return self.df.describe()

    def get_EDA_shape(self):
        return self.df.shape

    def get_EDA_value_counts(self):
        return self.df.value_counts()

    def get_EDA_isnull(self):
        return self.df.isnull()

    def get_EDA_dtypes(self):
        return self.df.dtypes

    def get_EDA_isna(self):
        return self.df.isna()

    def get_EDA_nunique(self):
        return self.df.nunique()

    def get_EDA_sort_dipl(self):
        return self.df.query("col1 == 'X'")

    def get_EDA_sort_bach(self):
        return self.df.query("col1 == 'Y'")

    def get_EDA_sort_by_line(self):
        return self.df.groupby(['col2', 'col1', 'col3']).agg(['count'])# groupby(['User Name', 'col2'])['col1'].size().reset_index(name='counts')

    '''
    Time series
    '''
    import matplotlib.pyplot as plt
    def get_time_series(self):
        df['Logon Time'] = pd.to_datetime(df['Logon Time'], errors='coerce')
        df['Year'] = df.index.dt.year
        df['month'] = df.index.dt.month
        df['day'] = df.inde.dt.day
        df['hour'] = df.index.dt.hour
        df['week'] = df.index.dt.week
        df['count'] = df['User Name']
        return df.groupby([df['Logon Time'].dt.year, df['Logon Time'].dt.month]).sum().plot.bar()
    plt.show()

...並從 main_count_stats.py 運行函數（從 df_count_stats.py 加載數據）：

from df_count_stats import df_load, df1_load
from formulas_count_stats import Data_load_compare_0, Data_load_compare_1

myData = Data_load_compare_0(df_load)
myData1 = Data_load_compare_1(df1_load)

EDA_stats_00_0 = myData.get_EDA_columns()
EDA_stats_01_0 = myData.get_EDA_nunique()
EDA_stats_02_0 = myData.get_EDA_shape()
EDA_stats_03_0 = myData.get_EDA_info()
EDA_stats_04_0 = myData.get_EDA_isna()
EDA_stats_05_0 = myData.get_EDA_isnull()
EDA_stats_06_0 = myData.get_EDA_describe()
EDA_stats_07_0 = myData.get_EDA_dtypes()
EDA_stats_08_0 = myData.get_EDA_sort_bach()
EDA_stats_09_0 = myData.get_EDA_sort_dipl()
EDA_stats_10_0 = myData.get_EDA_sort_by_line()
EDA_stats_11_0 = myData.get_time_series()

我收到此錯誤：

Traceback (most recent call last):
  File "C:/.../.../main_count_stats.py", line 25, in <module>
    EDA_stats_11_0 = myData.get_time_series()
  File "C:\...\...\...\formulas_count_stats.py", line 59, in get_time_series
    df['Year'] = df.index.dt.year
AttributeError: 'RangeIndex' object has no attribute 'dt'

我希望我嘗試將簡單的時間序列分析作為其他功能良好的 formulas_count_stats.py 的一部分進行整合。 顯然，事實並非如此。 我確實將索引更改為“to_datetime”格式。

我該如何解決這個問題？

Answer 1

或者您可以將索引轉換為Series並使用.dt ：

def get_time_series(self):
    self.df['Logon Time'] = pd.to_datetime(df['Logon Time'], errors='coerce')
    self.df[['Year', 'month', 'day', 'hour', 'week']] = (pd.Series(df.index)
                                                .dt.strftime('%Y-%m-%d-%H-%W')
                                                .str.split('-', expand=True).astype(int)).values
    self.df['count'] = df['User Name']
    return self.df.groupby([df['Logon Time'].dt.year, df['Logon Time'].dt.month]).sum().plot.bar()