basic lstm

import matplotlib.pyplot as plt
from datetime import datetime
import pandas as pd
import numpy as np
import requests

import silence_tensorflow.auto
from keras.models import Sequential
from keras.layers import Dense
from keras.layers import LSTM
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import mean_squared_error

# Dataset
if True:
    json = requests.get('https://api.kraken.com/0/public/OHLC?pair=BTCUSDT&interval=1440&since=1483228800').json()
    data = json['result']
    data = data[list(data.keys())[0]]
    df = pd.DataFrame(data, columns = ['date', 'open', 'high', 'low', 'close', 'vwap', 'volume', 'number_of_trades'])

    for x in df.columns:
        if df[x].dtypes == np.dtype(object):
            df[x] = df[x].astype(np.float32)

    df['date'] = pd.to_datetime(df['date'], unit='s')
    dataset = df['close'][-1000:]
else:
    from random import random
    dataset = [random() for x in range(100)]

look_back = int(len(dataset)/10)
print(dataset)

# Prepare data
dataset = np.array(dataset).reshape(-1, 1)
scaler = MinMaxScaler(feature_range=(0, 1))
dataset = scaler.fit_transform(dataset)

def split(dataset, p):
  train_size = int(len(dataset) * p)
  test_size = len(dataset) - train_size
  return dataset[0:train_size,:], dataset[train_size:len(dataset), :]

def create_dataset(_dataset, _look_back=1):
    data_x, data_y = [], []
    for i in range(len(_dataset) - _look_back - 1):
        a = _dataset[i:(i + _look_back), 0]
        data_x.append(a)
        data_y.append(_dataset[i + _look_back, 0])
    data_x = np.array(data_x)
    data_x = np.reshape(data_x, (data_x.shape[0], 1, data_x.shape[1]))
    return data_x, np.array(data_y)

train, test = split(dataset, 0.5)
train_x, train_y = create_dataset(train, look_back)
test_x, test_y = create_dataset(test, look_back)

# Model
model = Sequential()
model.add(LSTM(4, input_shape=(1, look_back)))
model.add(Dense(1))
model.compile(loss='mean_squared_error', optimizer='adam')
model.fit(train_x, train_y, epochs=1, batch_size=1)

train_predict = model.predict(train_x)
test_predict = model.predict(test_x)

# Transform result
train_predict = scaler.inverse_transform(train_predict)
train_y = scaler.inverse_transform([train_y])
test_predict = scaler.inverse_transform(test_predict)
test_y = scaler.inverse_transform([test_y])

# Plot
def to_plot(dataset, predict, start, end):
  predict_plot = np.empty_like(dataset)
  predict_plot[:, :] = np.nan
  predict_plot[start:end, :] = predict
  return predict_plot

train_predict_plot = to_plot(dataset, train_predict, look_back, len(train_predict) + look_back)
test_predict_plot = to_plot(dataset, test_predict, look_back + len(train_predict) + look_back + 1, len(dataset) - 1)
dates = pd.date_range(start=df['date'][0], periods=len(dataset))

plt.plot(dates, scaler.inverse_transform(dataset), label='dataset')
plt.plot(dates, train_predict_plot, label='train')
plt.plot(dates, test_predict_plot, label='test')
plt.gcf().autofmt_xdate()
plt.legend()
plt.show()