Time Series regression - MLP¶

In [1]:

# DAL ToolBox
# version 1.01.727

source("https://raw.githubusercontent.com/cefet-rj-dal/daltoolbox/main/jupyter.R")

#loading DAL
load_library("daltoolbox")

Loading required package: daltoolbox

Registered S3 method overwritten by 'quantmod':
  method            from
  as.zoo.data.frame zoo 


Attaching package: ‘daltoolbox’


The following object is masked from ‘package:base’:

    transform

Series for studying¶

In [2]:

data(sin_data)
ts <- ts_data(sin_data$y, 10)
ts_head(ts, 3)

A matrix: 3 × 10 of type dbl
t9	t8	t7	t6	t5	t4	t3	t2	t1	t0
0.0000000	0.2474040	0.4794255	0.6816388	0.8414710	0.9489846	0.9974950	0.9839859	0.9092974	0.7780732
0.2474040	0.4794255	0.6816388	0.8414710	0.9489846	0.9974950	0.9839859	0.9092974	0.7780732	0.5984721
0.4794255	0.6816388	0.8414710	0.9489846	0.9974950	0.9839859	0.9092974	0.7780732	0.5984721	0.3816610

In [3]:

library(ggplot2)
plot_ts(x=sin_data$x, y=sin_data$y) + theme(text = element_text(size=16))

data sampling¶

In [4]:

samp <- ts_sample(ts, test_size = 5)
io_train <- ts_projection(samp$train)
io_test <- ts_projection(samp$test)

data preprocessing¶

In [5]:

preproc <- ts_norm_gminmax()

Model training¶

In [6]:

model <- ts_mlp(ts_norm_gminmax(), input_size=4, size=4, decay=0)
model <- fit(model, x=io_train$input, y=io_train$output)

Evaluation of adjustment¶

In [7]:

adjust <- predict(model, io_train$input)
adjust <- as.vector(adjust)
output <- as.vector(io_train$output)
ev_adjust <- evaluate(model, output, adjust)
ev_adjust$mse

1.11558705263708e-05

Prediction of test¶

In [8]:

prediction <- predict(model, x=io_test$input[1,], steps_ahead=5)
prediction <- as.vector(prediction)
output <- as.vector(io_test$output)
ev_test <- evaluate(model, output, prediction)
ev_test

$values

0.412118485241757
0.173889485380434
-0.0751511204618093
-0.319519193622274
-0.54402111088937

$prediction

0.417930383863916
0.184270190591257
-0.0590592888553944
-0.300328614290653
-0.522706739711133

$smape

0.0827306326137497

$mse

0.000244613000904878

$R2

0.997887264876547

$metrics

A data.frame: 1 × 3
mse	smape	R2
<dbl>	<dbl>	<dbl>
0.000244613	0.08273063	0.9978873

Plot results¶

In [9]:

yvalues <- c(io_train$output, io_test$output)
plot_ts_pred(y=yvalues, yadj=adjust, ypre=prediction) + theme(text = element_text(size=16))

In [ ]: