aml_make_model: Function to train Deep Learning regression model for a single asset

Description

Function is training h2o deep learning model to match future prices of the asset to the indicator pattern. Main idea is to be able to predict future prices by solely relying on the recently retrieved indicator pattern. This is to mimic traditional algorithmic systems based on the indicator rule attempting to automate optimization process with AI.

Deep learning model structure is obtained from the 8 random combinations of neurons within 3 hidden layers of the network, the most accurate model configuration will be automatically selected

In addition, the function will check if there is a need to update the model. To do that function will check results of the function aml_test_model.R.

Usage

aml_make_model(
  symbol,
  timeframe,
  path_model,
  path_data,
  force_update = FALSE,
  num_nn_options = 24
)

Arguments

symbol

Character symbol of the asset for which to train the model

timeframe

Data timeframe e.g. 1 min

path_model

Path where the models shall be stored

path_data

Path where the aggregated historical data is stored, if exists in rds format

force_update

Boolean, by setting this to TRUE function will generate new model (useful after h2o engine update)

num_nn_options

Integer, value from 1 to 20 or more. Used to change number of variants of the random neural network structures

Value

Function is writing a file object with the best Deep Learning Regression model

Details

Function is using the dataset prepared by the function aml_collect_data.R. Function will start to train the model as soon as there are more than 100 rows in the dataset

Examples

Run this code

# NOT RUN {
# }
# NOT RUN {

library(dplyr)
library(readr)
library(h2o)
library(lazytrade)
library(lubridate)

path_model <- normalizePath(tempdir(),winslash = "/")
path_data <- normalizePath(tempdir(),winslash = "/")

ind = system.file("extdata", "AI_RSIADXUSDJPY60.csv",
                  package = "lazytrade") %>% read_csv(col_names = FALSE)

ind$X1 <- ymd_hms(ind$X1)


# data transformation using the custom function for one symbol
aml_collect_data(indicator_dataset = ind,
                 symbol = 'USDJPY',
                 timeframe = 60,
                 path_data = path_data)

# dataset will be written to the temp directory

# start h2o engine
h2o.init(nthreads = 2)


# performing Deep Learning Regression using the custom function
aml_make_model(symbol = 'USDJPY',
               timeframe = 60,
               path_model = path_model,
               path_data = path_data,
               force_update=FALSE,
               num_nn_options = 2)

# stop h2o engine
h2o.shutdown(prompt = FALSE)

#set delay to insure h2o unit closes properly before the next test
Sys.sleep(5)

# }
# NOT RUN {


# }