Work with a custom dataset; not the reuters dataset about lda HOT 4 CLOSED

import numpy as np
import lda
import csv

matrix = np.loadtxt(open("data.csv", "rb"), delimiter=",", skiprows=1).astype("int64")
X = matrix[:,1:] # X is training data, matrix is data with doc id

model = lda.LDA(n_topics=100, n_iter=1000, random_state=1)
model.fit(X)

doc_ids = matrix[:,0]
doc_topic = model.doc_topic_

with open('doc_topics.csv', 'w', newline='') as csvfile:
    writer = csv.writer(csvfile, delimiter=',', quoting=csv.QUOTE_MINIMAL)
    writer.writerow(["doc_id", "topic_id"])
    for i, doc_id in enumerate(doc_ids):
        writer.writerow([doc_id, doc_topic[i].argmax()])

>>> import lda.datasets
>>> X = lda.datasets.load_reuters()
>>> X
array([[1, 0, 1, ..., 0, 0, 0],
       [7, 0, 2, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       ...,
       [1, 0, 1, ..., 0, 0, 0],
       [1, 0, 1, ..., 0, 0, 0],
       [1, 0, 1, ..., 0, 0, 0]], dtype=int32)
>>> X[0]
array([1, 0, 1, ..., 0, 0, 0], dtype=int32)
>>> len(X[0])
4258
>>> X[0][0]
1