VAE:输入数据形状(多变量)

时间:2019-08-05 02:07:12

标签: python pandas numpy keras deep-learning

使用VAE模型时,输入数据为(12,1)形状。

  • 数据具有季节性特征。因此,我从一月到十二月做了每一列。

然后我想将温度和湿度数据作为变量来改进模型。

'''
1. My data format
'''
sample_df.head()

  area  year  month   id    usage
1   1   2009    1   11111   6670    
2   1   2009    2   11111   5746    

'''
2. After Pivoting
'''
sample_df.head()

  id      1M      2M      3M...        9M       10M      11M      12M                                               
11111   6670.00 5746.00 4608.00...   4962.00  6987.00   8051.00 8325.00
...


'''
3. My Initial model Structure
'''
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
encoder_input (InputLayer)      (None, 12, 1)        0                                            
__________________________________________________________________________________________________
conv1d_1 (Conv1D)               (None, 10, 4)        16          encoder_input[0][0]              
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 10, 4)        16          conv1d_1[0][0]                   
__________________________________________________________________________________________________
leaky_re_lu_1 (LeakyReLU)       (None, 10, 4)        0           batch_normalization_1[0][0]      
__________________________________________________________________________________________________
conv1d_2 (Conv1D)               (None, 8, 8)         104         leaky_re_lu_1[0][0]              
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 8, 8)         32          conv1d_2[0][0]                   
__________________________________________________________________________________________________
leaky_re_lu_2 (LeakyReLU)       (None, 8, 8)         0           batch_normalization_2[0][0]      
__________________________________________________________________________________________________
conv1d_3 (Conv1D)               (None, 6, 16)        400         leaky_re_lu_2[0][0]              
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 6, 16)        64          conv1d_3[0][0]                   
__________________________________________________________________________________________________
leaky_re_lu_3 (LeakyReLU)       (None, 6, 16)        0           batch_normalization_3[0][0]      
__________________________________________________________________________________________________
conv1d_4 (Conv1D)               (None, 6, 1)         17          leaky_re_lu_3[0][0]              
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 6, 1)         4           conv1d_4[0][0]                   
__________________________________________________________________________________________________
leaky_re_lu_4 (LeakyReLU)       (None, 6, 1)         0           batch_normalization_4[0][0]      
__________________________________________________________________________________________________
flatten_1 (Flatten)             (None, 6)            0           leaky_re_lu_4[0][0]              
__________________________________________________________________________________________________
z_mean (Dense)                  (None, 2)            14          flatten_1[0][0]                  
__________________________________________________________________________________________________
z_log_var (Dense)               (None, 2)            14          flatten_1[0][0]                  
__________________________________________________________________________________________________
z (Lambda)                      (None, 2)            0           z_mean[0][0]                     
                                                                 z_log_var[0][0]                  
__________________________________________________________________________________________________
decoder (Model)                 (None, 12, 1)        777         z[0][0]                          
==================================================================================================

对于每个ID,每个月都有温度和湿度数据,如何在VAE模型中将这些数据输入到输入数据中?

0 个答案:

没有答案