convert time series data set to supervised for deep learining

Question

convert time series data set to supervised for deep learining

Almas Co

2022年6月1日 13:31

I have dataset like

so I want to use that for prediction of time series with deep learning. I have this function to make it supervised

def to_supervised(train,n_input,n_out):
#falten data
data=train
X,y=list(),list()
in_start=0
for _ in range(len(data)):
    in_end=in_start+ n_input
    out_end=in_end + n_out
    if out_end=len(data):
        x_input=data[ in_start:in_end,0]
        x_input=x_input.reshape((len(x_input)))
        X.append(x_input)
        y.append(data[in_end:out_end,0])
    in_start+=1
return array(X), array(y)

I am not sure about functionality of this function. Do you have replacemment for this function?

Topic dataset

Category Data Science

Adept · Accepted Answer · 2022年6月1日 13:31

According to our comments, your objective is to "predict any issued according to the previous issues"

The best way to include time series data in such a dataset is to include variable crawling through the past dates.

For example :

Value of the latest known records
Value of the 7 latest known records
Variance of the 7 latest known records
Min, Max, etc

Example : If you want to predict in 2022-06-01, you only know data before this date, so creating variables taking past information like that can be a good idea. The biggest problem will be adapting your period to your problem (is there any gain in information if you look a few years before ?) and to handle your first cases (what do you put in the variable "Value of the 7 latest known 'issues'" if you only know 2 past records for example).

convert time series data set to supervised for deep learining

About