import torch
import sys
import torch.nn.functional as F
from torch import nn
from sklearn.metrics import accuracy_score, mean_squared_error
import numpy as np
import pandas as pd
from sacred import Experiment
from sacred.observers import FileStorageObserver

np.set_printoptions(suppress=False)

ex = Experiment("ium_s434766", interactive=False, save_git_info=False)
ex.observers.append(FileStorageObserver("ium_s434766/my_runs"))

@ex.config
def my_config():
    num_epochs = 15
    batch_size = 16
    learning_rate = 0.001

class LogisticRegressionModel(nn.Module):
    def __init__(self, input_dim, output_dim):
        super(LogisticRegressionModel, self).__init__()
        self.linear = nn.Linear(input_dim, output_dim)
        self.sigmoid = nn.Sigmoid()
    def forward(self, x):
        out = self.linear(x)
        return self.sigmoid(out)

@ex.capture
def train(num_epochs, batch_size, learning_rate, _log):
    data_train = pd.read_csv("data_train.csv")
    data_test  = pd.read_csv("data_test.csv")
    FEATURES   =  ['age','hypertension','heart_disease','ever_married', 'avg_glucose_level', 'bmi']

    x_train = data_train[FEATURES].astype(np.float32)
    y_train = data_train['stroke'].astype(np.float32)

    x_test = data_test[FEATURES].astype(np.float32)
    y_test = data_test['stroke'].astype(np.float32)

    fTrain = torch.from_numpy(x_train.values)
    tTrain = torch.from_numpy(y_train.values.reshape(2945,1))

    fTest= torch.from_numpy(x_test.values)
    tTest = torch.from_numpy(y_test.values)


    input_dim = 6
    output_dim = 1
    info_params = "Batch size = " + str(batch_size) + " Epochs = " + str(num_epochs)
    _log.info(info_params)
    model = LogisticRegressionModel(input_dim, output_dim)

    criterion = torch.nn.BCELoss(reduction='mean') 
    optimizer = torch.optim.SGD(model.parameters(), lr = learning_rate)

    for epoch in range(num_epochs):
        # print ("Epoch #",epoch)
        model.train()
        optimizer.zero_grad()
        # Forward pass
        y_pred = model(fTrain)
        # Compute Loss
        loss = criterion(y_pred, tTrain)
        # print(loss.item())
        # Backward pass
        loss.backward()
        optimizer.step()
        
    info_loss = "Last loss = " + str(loss.item())
    _log.info(info_loss)
    y_pred = model(fTest)
    # print("predicted Y value: ", y_pred.data)

    torch.save(model.state_dict(), 'stroke.pth')


@ex.automain
def my_main(num_epochs, batch_size, learning_rate, _run):
    train()

r = ex.run()
ex.add_artifact("stroke.pth")