model.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""
LSTM models used for predicting
"""

import os
from os.path import join as pj

from keras.layers import LSTM, Dense, Activation, Dropout
from keras.models import Sequential


base_dir = os.path.dirname(__file__)
weights_dir = pj(base_dir, 'weights')


def get_model(model=1, input_size=16, output_size=16, maxlen=90):
    """Creates as keras LSTM model

    Parameters
    ----------
    model: int
        Number of layers of the model
    input_size: int
        Number of input symptoms to take into account
    output_size: int
        Number of outputs symptoms to predict
    maxlen: int
        Number of days in the past that are required to generate next day prediction

    Returns
    -------
    model: keras.model
        Model used for training and/or prediction
    """
    assert model in [1, 2]
    if model == 1:
        model = Sequential()
        model.add(LSTM(128, input_shape=(maxlen, input_size)))
        model.add(Dropout(0.5))
        model.add(Dense(output_size))
        model.add(Activation('sigmoid'))

    elif model == 2:
        model = Sequential()
        model.add(LSTM(256, input_shape=(maxlen, input_size), return_sequences=True))
        model.add(Dropout(0.5))
        model.add(LSTM(256))
        model.add(Dropout(0.6))
        model.add(Dense(output_size))
        model.add(Activation('sigmoid'))

    return model


def get_weight_path(model, input_size, output_size, maxlen):
    """Creates the model name based on the parameters

    Parameters
    ----------
    model: int
        Number of layers of the model
    input_size: int
        Number of input symptoms to take into account
    output_size: int
        Number of outputs symptoms to predict
    maxlen: int
        Number of days in the past that are required to generate next day prediction

    Returns
    -------
    :string
        Name of the model
    """
    template = "lstm_{}_layers_{}_input_size_{}_output_size_{}_maxlen.hdf5"
    return pj(weights_dir, template.format(model, input_size, output_size, maxlen))