Machine learning in trading: theory, models, practice and algo-trading - page 3404

New comment
 
Aleksey Vyazmikin #:

As I realise there is a lot of randomness in balancing, I got this result.

try it

 
Aleksey Vyazmikin #:

Well, judging by your code you have different expectations - not what I need :)

What did I do there? =)

 
mytarmailS #:

I need to save the indices of the excluded predictors, not the selected predictors! I need to reduce the index by one. Can you correct your code?

 
Aleksey Vyazmikin #:

I need to save the indices of the excluded predictors, not the selected predictors! I need to reduce the index by one. Can you correct your code?


which(original_colum_names %in% colnames(data)[-bad_colums])

which(original_colum_names %in% colnames(data)[  bad_colums])-1
 

And in the feature selection code, you have to

best_colums_idx <- which(original_colum_names %in% ex$support.vars)

best_colums_idx <- which(! original_colum_names %in% ex$support.vars)-1
 

Question for connoisseurs.

First of all, introductory information. This is a complex plane, the frequency (angle) and attenuation (amplitude) of an exponential "signal" are determined on it. The unit circle separates damped and increasing signals (by the way, in econometrics they try to pick up AR model k-ty so that its roots lie on this circle, like a sinusoid should generate). Red and blue points are roots of 2 different models for several different runs. If there is a signal, the blue points should be outside the circle and they should correspond to the red points inside the circle.

This is an exponent with different noise levels (artificial series).

Low noise 10 times less than the signal 2 times less than the signal Equal to the signal


And these are the clock increments. The distribution is close to normal.

Attention question. As you can see, there is no obvious signal on the sentinels, but there is clustering of noise. What can it be?


PS And what have you achieved in the fight against randomness?

 
Rorschach #:

Attention question. As you can see, there is no obvious signal on the hourly, but there is clustering of noise. What can it be

I'm no expert, but I think the answer is obvious. It sounds something like this - "visualisation".

 
Rorschach #:

Attention question. As you can see, there is no obvious signal on the hourly, but there is noise clustering . What can it be?

volatility clustering mb

 
mytarmailS #:


Write a complete piece of the last code, please. I don't know much about R and I don't quite understand where to insert code fragments

I did it like this.

# Загрузка данных из CSV
data <- read.csv("E:\\FX\\MT5_CB\\MQL5\\Files\\00_Standart_50\\Setup\\train.csv", sep = ";")
data <- data[, !(names(data) %in% c("Time", "Target_P", "Target_100", "Target_100_Buy", "Target_100_Sell"))]


bad_colums <- caret::findLinearCombos(data[1:500,])$remove

ncol(data)         # количество всех колонок 2408
length(bad_colums) # кличество линейно зависимых 1908


clear_data <- cbind.data.frame(
  target, 
  data[,-bad_colums]
)


which(original_colum_names %in% colnames(data)[-bad_colums])
which(original_colum_names %in% colnames(data)[ bad_colums])-1

remaining_columns_indices <- which(original_colum_names %in% colnames(data)[,-bad_colums])
remaining_columns_indices_minus1 <- which(original_colum_names %in% colnames(data)[bad_colums]) - 1

write.csv(remaining_columns_indices_minus1, "E:\\FX\\MT5_CB\\MQL5\\Files\\00_Standart_50\\Setup\\Оставшиеся_предикторы_00.csv", row.names = FALSE)

I got an error

> remaining_columns_indices <- which(original_colum_names %in% colnames(data)[,-bad_colums])
Error in colnames(data)[, -bad_colums] : incorrect number of dimensions
 
Rorschach #:

Attention question. As you can see, there is no obvious signal on the hourly, but there is noise clustering . What can it be?

Without understanding the context, when I see such phrases, I immediately start thinking about the dependence on the time of day. The lower (starting from H1) the timeframe, the weaker the signal is on the increments.

1...339733983399340034013402340334043405340634073408340934103411...3626
New comment