how to implement tensorflow's next_batch for own data

In order to shuffle and sampling each mini-batch, the state whether a sample has been selected inside the current epoch should also be considered. Here is an implementation which use the data in the above answer.

import numpy as np 

class Dataset:

def __init__(self,data):
    self._index_in_epoch = 0
    self._epochs_completed = 0
    self._data = data
    self._num_examples = data.shape[0]
    pass


@property
def data(self):
    return self._data

def next_batch(self,batch_size,shuffle = True):
    start = self._index_in_epoch
    if start == 0 and self._epochs_completed == 0:
        idx = np.arange(0, self._num_examples)  # get all possible indexes
        np.random.shuffle(idx)  # shuffle indexe
        self._data = self.data[idx]  # get list of `num` random samples

    # go to the next batch
    if start + batch_size > self._num_examples:
        self._epochs_completed += 1
        rest_num_examples = self._num_examples - start
        data_rest_part = self.data[start:self._num_examples]
        idx0 = np.arange(0, self._num_examples)  # get all possible indexes
        np.random.shuffle(idx0)  # shuffle indexes
        self._data = self.data[idx0]  # get list of `num` random samples

        start = 0
        self._index_in_epoch = batch_size - rest_num_examples #avoid the case where the #sample != integar times of batch_size
        end =  self._index_in_epoch  
        data_new_part =  self._data[start:end]  
        return np.concatenate((data_rest_part, data_new_part), axis=0)
    else:
        self._index_in_epoch += batch_size
        end = self._index_in_epoch
        return self._data[start:end]

dataset = Dataset(np.arange(0, 10))
for i in range(10):
    print(dataset.next_batch(5))

the output is:

[2 8 6 3 4]
[1 5 9 0 7]
[1 7 3 0 8]
[2 6 5 9 4]
[1 0 4 8 3]
[7 6 2 9 5]
[9 5 4 6 2]
[0 1 8 7 3]
[9 7 8 1 6]
[3 5 2 4 0]

the first and second (3rd and 4th,…) mini-batch correspond to one whole epoch..

I use Anaconda and Jupyter.
In Jupyter if you run ?mnist you get:
File: c:programdataanaconda3libsite-packagestensorflowcontriblearnpythonlearndatasetsbase.py
Docstring: Datasets(train, validation, test)

In folder datesets you shall find mnist.py which contains all methods including next_batch.

If you would not like to get shape mismatch error in your tensorflow session run
then use the below function instead of the function provided in the first solution above (https://stackoverflow.com/a/40995666/7748451) –

def next_batch(num, data, labels):

    '''
    Return a total of `num` random samples and labels. 
    '''
    idx = np.arange(0 , len(data))
    np.random.shuffle(idx)
    idx = idx[:num]
    data_shuffle = data[idx]
    labels_shuffle = labels[idx]
    labels_shuffle = np.asarray(labels_shuffle.values.reshape(len(labels_shuffle), 1))

    return data_shuffle, labels_shuffle

The answer which is marked up above I tried the algorithm by that algorithm I am not getting results so I searched on kaggle and I saw really amazing algorithm which worked really well. Best result try this. In below algorithm **Global variable takes the input you declared above in which you read your data set.**

epochs_completed = 0
index_in_epoch = 0
num_examples = X_train.shape[0]
    # for splitting out batches of data
def next_batch(batch_size):

    global X_train
    global y_train
    global index_in_epoch
    global epochs_completed

    start = index_in_epoch
    index_in_epoch += batch_size

    # when all trainig data have been already used, it is reorder randomly    
    if index_in_epoch > num_examples:
        # finished epoch
        epochs_completed += 1
        # shuffle the data
        perm = np.arange(num_examples)
        np.random.shuffle(perm)
        X_train = X_train[perm]
        y_train = y_train[perm]
        # start next epoch
        start = 0
        index_in_epoch = batch_size
        assert batch_size <= num_examples
    end = index_in_epoch
    return X_train[start:end], y_train[start:end]

Yet another implementation:

from typing import Tuple
import numpy as np

class BatchMaker(object):
    def __init__(self, feat: np.array, lab: np.array) -> None:
        if len(feat) != len(lab):
            raise ValueError("Expected feat and lab to have the same number of samples")
        self.feat = feat
        self.lab = lab
        self.indexes = np.arange(len(feat))
        np.random.shuffle(self.indexes)
        self.pos = 0

    # "BatchMaker, BatchMaker, make me a batch..."
    def next_batch(self, batch_size: int) -> Tuple[np.array, np.array]:
        if self.pos + batch_size > len(self.feat):
            np.random.shuffle(self.indexes)
            self.pos = 0
        batch_indexes = self.indexes[self.pos: self.pos + batch_size]
        self.pos += batch_size
        return self.feat[batch_indexes], self.lab[batch_indexes]