After breaking the series into 3 parts and then compute mean and variance for each part. I need to understand what is the acceptable difference between the 3 parts to say this series is stationary?

This is the output for 2 series:


mean1= 19.079044     mean2= 21.812044    mean3= 18.575845   
variance1= 1.437688  variance2= 1.151551     variance3= 0.547118


mean1= 0.207178  mean2= 0.146585     mean3= -0.333707   
variance1= 0.999076  variance2= 0.950360     variance3= 0.974912

I think that the first series is stationary because the difference between the mean is very little, but what about the second series?

This is my code:

data = pd.read_csv("C://r3800.txt")  
values = data.values
# getting the count to split the dataset into 3
parts = int(len(values)/3)
# splitting the data into three parts
part_1, part_2, part_3 = values[0:parts], values[parts:(
    parts*2)], values[(parts*2):(parts*3)] 
# calculating the mean of the separated three
# parts of data individually.
mean_1, mean_2, mean_3 = part_1.mean(), part_2.mean(), part_3.mean() 
# calculating the variance of the separated
# three parts of data individually.
var_1, var_2, var_3 = part_1.var(), part_2.var(), part_3.var() 
# printing the mean of three groups
print('mean1= %f\t mean2= %f\t mean3= %f \t' % (mean_1, mean_2, mean_3)) 
# printing the variance of three groups
print('variance1= %f\t variance2= %f \t variance3= %f' % (var_1, var_2, var_3))
  • $\begingroup$ On what basis did you split the series? How long is each part of each series? $\endgroup$
    – whuber
    Aug 30, 2022 at 18:49
  • $\begingroup$ I will share the code $\endgroup$
    – ayla
    Aug 30, 2022 at 18:50
  • $\begingroup$ Wouldn't it suffice to say you're splitting them into thirds, with a little overlap at the splits? Again, how long are these series? That's a crucial piece of information. $\endgroup$
    – whuber
    Aug 30, 2022 at 18:56
  • $\begingroup$ The length is 3800 values $\endgroup$
    – ayla
    Aug 30, 2022 at 18:58
  • $\begingroup$ That indicates the variances are tiny indeed. But a decision ultimately has to rest on how autocorrelated these series are, too. $\endgroup$
    – whuber
    Aug 30, 2022 at 19:01


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.