0

'有人可以分解这段代码吗？如何在没有循环的情况下设置 groupby 和 set_index'

import pmdarima as pm
from pmdarima.model_selection import train_test_split

def MAPE(y_orig, y_pred):
    diff = y_orig - y_pred
    MAPE = np.mean((abs(y_orig - y_pred)/y_orig)  * 100.)
    return MAPE

Can I use this model without using the **def function** and instead of using MAPE calculations in the function, Can I use the MAPE calculations after running the model (separately)

def XYZ_analysis(df):
    XYZ_list = []
    for group, value in df.groupby(['CustomerID']):
        grouped_df = (
                value.loc[:, ['Timeperiod','Quantity']].set_index('Timeperiod')      
        )

        # Using a simple Seasonal ARIMA model to highlight the idea, in the actual world, the model has to best fit the data
        train, test = train_test_split(grouped_df, train_size=18)

        model = pm.auto_arima(train, seasonal=True, m=4)

        forecasts = model.predict(test.shape[0])  # predict N steps into the future
        mape = MAPE(test.values.reshape(1,-1)[0], forecasts)
        XYZ_list.append([group, mape])
     
    XYZ_group = pd.DataFrame(XYZ_list, columns = ['CustomerID', 'MAPE'])
    XYZ_group = XYZ_group.sort_values(by=['MAPE'], ascending=True)
    
    ## Ranking by forecastability
    XYZ_group["Rank"] = XYZ_group['MAPE'].rank(ascending = True)
    XYZ_group["Forecastability"] = ' '
    
    ## Checking the Importance of the Customers and Categorising into class A,B,C and splitting based on 20-30-50
    XYZ_group['Forecastability'][0: int(0.2 * XYZ_group['Rank'].max())] = 'X'
    XYZ_group['Forecastability'][int(0.2 * XYZ_group['Rank'].max()) : int(0.5 * XYZ_group['Rank'].max())] = 'Y'
    XYZ_group['Forecastability'][int(0.5 * XYZ_group['Rank'].max()): ] = 'Z'
    
    return XYZ_group

' 代码来自此页面https://github.com/adib0073/Importance_Forecastability_Segmentation/blob/master/ABC-XYZ%20Analysis.ipynb '

'谢谢'

4

0 回答 0