pre-processing

import pandas as pd
from scipy import sparse 
import numpy as np
import scipy.sparse

data=pd.read_csv('C:\\Users\\harsh\\Downloads\\training (2).csv')
vocab = pd.read_csv("C:\\Users\\harsh\\Downloads\\vocabularyc.csv")
classc= pd.read_csv("C:\\Users\\harsh\\Downloads\\newsgrouplabels.csv")
classc.columns=['id','group']
classc=classc.to_numpy()
vocab_np = vocab.to_numpy()
classc2=np.delete(classc,0,axis=1)
dictlist = [dict() for x in range(5)]
arr = data.to_numpy()
x_train_sparse=sparse.csr_matrix(arr)
x_train_sparse_length = arr.shape[0]
x_lastCol = arr[:,-1]
dltlastcol = np.delete(arr, -1, axis=1)    
dltlastcol1 = np.delete(dltlastcol,0,axis=1) 
x_train_sparse=sparse.csr_matrix(dltlastcol1)
x_train_sparse_length = dltlastcol1.shape[0]
res_temp=vocab_np[0]
clss_temp=classc2[x_lastCol[4]]
rows,cols = x_train_sparse.nonzero()

dict_rows_as_docs = [dict() for x in range(x_train_sparse_length)]

for row,col in zip(rows,cols):
    voc_temp=vocab_np[col]
    class_value= classc2[x_lastCol[row]-1]
    dict_rows_as_docs[row][voc_temp[0]]=(x_train_sparse[row,col],class_value[0])
dict_rows_as_docs