scripts/model.py

"""
Attribution: https://github.com/AIPI540/AIPI540-Deep-Learning-Applications/

Jon Reifschneider
Brinnae Bent 

"""

import os
import urllib.request
import zipfile
import json
import pandas as pd
import time
import torch
import numpy as np
import pandas as pd
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torch.utils.data import DataLoader, TensorDataset
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
from sklearn.preprocessing import LabelEncoder




def prep_dataloaders(X_train,y_train,X_val,y_val,batch_size):
    # Convert training and test data to TensorDatasets
    trainset = TensorDataset(torch.from_numpy(np.array(X_train)).long(),
                            torch.from_numpy(np.array(y_train)).float())
    valset = TensorDataset(torch.from_numpy(np.array(X_val)).long(),
                            torch.from_numpy(np.array(y_val)).float())

    # Create Dataloaders for our training and test data to allow us to iterate over minibatches
    trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True)
    valloader = torch.utils.data.DataLoader(valset, batch_size=batch_size, shuffle=False)

    return trainloader, valloader


class NNColabFiltering(nn.Module):

    def __init__(self, n_playlists, n_artists, embedding_dim_users, embedding_dim_items, n_activations, rating_range):
        super().__init__()
        self.user_embeddings = nn.Embedding(num_embeddings=n_playlists,embedding_dim=embedding_dim_users)
        self.item_embeddings = nn.Embedding(num_embeddings=n_artists,embedding_dim=embedding_dim_items)
        self.fc1 = nn.Linear(embedding_dim_users+embedding_dim_items,n_activations)
        self.fc2 = nn.Linear(n_activations,1)
        self.rating_range = rating_range

    def forward(self, X):
        # Get embeddings for minibatch
        embedded_users = self.user_embeddings(X[:,0])
        embedded_items = self.item_embeddings(X[:,1])
        # Concatenate user and item embeddings
        embeddings = torch.cat([embedded_users,embedded_items],dim=1)
        # Pass embeddings through network
        preds = self.fc1(embeddings)
        preds = F.relu(preds)
        preds = self.fc2(preds)
        # Scale predicted ratings to target-range [low,high]
        preds = torch.sigmoid(preds) * (self.rating_range[1]-self.rating_range[0]) + self.rating_range[0]
        return preds

def train_model(model, criterion, optimizer, dataloaders, device, num_epochs=5, scheduler=None):

    model = model.to(device) # Send model to GPU if available
    since = time.time()

    costpaths = {'train':[],'val':[]}

    for epoch in range(num_epochs):
        print('Epoch {}/{}'.format(epoch, num_epochs - 1))
        print('-' * 10)

        # Each epoch has a training and validation phase
        for phase in ['train', 'val']:
            if phase == 'train':
                model.train()  # Set model to training mode
            else:
                model.eval()   # Set model to evaluate mode

            running_loss = 0.0

            # Get the inputs and labels, and send to GPU if available
            index = 0
            for (inputs,labels) in dataloaders[phase]:
                inputs = inputs.to(device)
                labels = labels.to(device)

                # Zero the weight gradients
                optimizer.zero_grad()

                # Forward pass to get outputs and calculate loss
                # Track gradient only for training data
                with torch.set_grad_enabled(phase == 'train'):
                    outputs = model.forward(inputs).view(-1)
                    loss = criterion(outputs, labels)

                    # Backpropagation to get the gradients with respect to each weight
                    # Only if in train
                    if phase == 'train':
                        loss.backward()
                        # Update the weights
                        optimizer.step()

                # Convert loss into a scalar and add it to running_loss
                running_loss += np.sqrt(loss.item()) * labels.size(0)
                print(f'\r{running_loss} {index} {(index / len(dataloaders[phase]))*100:.2f}%', end='')
                index +=1

            # Step along learning rate scheduler when in train
            if (phase == 'train') and (scheduler is not None):
                scheduler.step()

            # Calculate and display average loss and accuracy for the epoch
            epoch_loss = running_loss / len(dataloaders[phase].dataset)
            costpaths[phase].append(epoch_loss)
            print('\n{} loss: {:.4f}'.format(phase, epoch_loss))

    time_elapsed = time.time() - since
    print('Training complete in {:.0f}m {:.0f}s'.format(
        time_elapsed // 60, time_elapsed % 60))

    return costpaths


if __name__ == '__main__':
    artists = pd.read_csv(os.getcwd() + '/data/processed/playlists.csv')
    X = artists.loc[:,['playlist_id','artist_album_id',]]
    y = artists.loc[:,'song_percent']

    # Split our data into training and test sets
    X_train, X_val, y_train, y_val = train_test_split(X,y,random_state=0, test_size=0.2)
    batchsize = 64
    trainloader,valloader = prep_dataloaders(X_train,y_train,X_val,y_val,batchsize)

    dataloaders = {'train':trainloader, 'val':valloader}
    n_users = X.loc[:,'playlist_id'].max()+1
    n_items = X.loc[:,'artist_album_id'].max()+1
    model = NNColabFiltering(n_users,n_items,embedding_dim_users=50, embedding_dim_items=50, n_activations = 100,rating_range=[0.,1.])
    criterion = nn.MSELoss()
    lr=0.001
    n_epochs=10
    wd=1e-3
    optimizer = optim.Adam(model.parameters(), lr=lr, weight_decay=wd)
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

    cost_paths = train_model(model,criterion,optimizer,dataloaders, device,n_epochs, scheduler=None)


    # Save the entire model
    torch.save(model, os.getcwd() + '/models/recommender.pt')