一些Pandas的Trick

def generate_sample_data(): # creates a fake df for testing
    number_or_rows = 20
    num_cols = 7
    cols = list("ABCDEFG")
    df = pd.DataFrame(np.random.randint(1, 20, size = (number_or_rows, num_cols)), columns=cols)
    df.index = pd.util.testing.makeIntIndex(number_or_rows)
    return df

# check Trick 91 for an example
def generate_sample_data_datetime(): # creates a fake df for testing
    number_or_rows = 365*24
    num_cols = 2
    cols = ["sales", "customers"]
    df = pd.DataFrame(np.random.randint(1, 20, size = (number_or_rows, num_cols)), columns=cols)
    df.index = pd.util.testing.makeDateIndex(number_or_rows, freq="H")
    return df

日期时间类型接口一览表（衍生特征可以糊好多）

df = generate_sample_data_datetime().reset_index()
df = df.sample(500)
df["Year"] = df["index"].dt.year
df["Month"] = df["index"].dt.month
df["Day"] = df["index"].dt.day
df["Hour"] = df["index"].dt.hour
df["Minute"] = df["index"].dt.minute
df["Second"] = df["index"].dt.second
df["Nanosecond"] = df["index"].dt.nanosecond
df["Date"] = df["index"].dt.date
df["Time"] = df["index"].dt.time
df["Time_Time_Zone"] = df["index"].dt.timetz
df["Day_Of_Year"] = df["index"].dt.dayofyear
df["Week_Of_Year"] = df["index"].dt.weekofyear
df["Week"] = df["index"].dt.week
df["Day_Of_week"] = df["index"].dt.dayofweek
df["Week_Day"] = df["index"].dt.weekday
df["Week_Day_Name"] = df["index"].dt.weekday_name
df["Quarter"] = df["index"].dt.quarter
df["Days_In_Month"] = df["index"].dt.days_in_month
df["Is_Month_Start"] = df["index"].dt.is_month_start
df["Is_Month_End"] = df["index"].dt.is_month_end
df["Is_Quarter_Start"] = df["index"].dt.is_quarter_start
df["Is_Quarter_End"] = df["index"].dt.is_quarter_end
df["Is_Leap_Year"] = df["index"].dt.is_leap_year