Trying to use Keras Dataloader for timeseries forecasting

In trying to create a time-series forecasting model, the main aim is to send data of product id in a singularity. For example, while using Keras.dataloader only data of product id 'mp000000000000005' should be sent and not any other data.

Below is the data I am using:

product_id         rpt_date index  total_views   total_cart_adds    total_order_unit    total_gmv   score
mp000000000000005   2022-07-24     5                  0                   0               0.0    0.584963
mp000000000000005   2022-07-25     1                  0                   0               0.0    1.736966
mp000000000000005   2022-07-26     0                  0                   0               0.0    0.000000
mp000000000001108   2022-07-10     4                  0                   0               0.0  0.263034 
mp000000000001108   2022-07-11     1                  0                   0               0.0  1.736966 
mp000000000001108   2022-07-12     0                  0                   0               0.0  0.000000  
mp000000000000055   2022-07-21     0                  0                   0               0.0  0.000000
mp000000000000055   2022-07-22     3                  0                   0               0.0  0.152003
mp000000000000055   2022-07-23     0                  0                   0               0.0  0.000000 
 

The requirement is that I want to send data of n days of the 1 product id at a time and then do the same for all the other product ids in the dataset.

This is the Code for DataLoader

class DataLoader(Sequence):
def __init__(self,X,y,batch_size=1024):
    self.X=X
    self.y=y
    self.batch_size=batch_size
def __len__(self):
    return len(self.y)
def __getitem__(self,index):
    #Code is mentioned below
    def create_dataset(dataset, look_back=1):
         dataX, dataY = [], []
            for p in range(len(dataset)-(look_back-1)):
                 a = dataset[p:(p+look_back), 0]
                  dataX.append(a)
                 dataY.append(dataset[p + look_back, 0])
         return np.array(dataX), np.array(dataY)
def __len__(self):
    return len(self.X)//self.batch_size