I am wanting to create an LSTM machine learning model in python using numpy and pandas, the model is for a music recommendation engine which takes in a streaming history of a specific length with Spotify music features. What is the most efficient way in tackling this in python using OOP, I want to compare it to a pretrained model to see if it performs better when trained specifically for this. I am relatively new in this field, it is for a school project.

# Import libraries
import numpy as np
import pandas as pd


class Network:
    def __init__(self):
        """
        Neural network for machine learning algorithm to form suggestions
        """
        # Import training data
        self.data = np.array(pd.read_csv('train_clean.csv'))
        # Fetch size of data array
        self.m, self.n = self.data.shape
        # Randomises order of training data
        np.random.shuffle(self.data)


net = Network()

This is the class layout I have started with, any ideas for how to tackle this would be greatly appreciated.