Randomness of time series

[drfg2008]

I have a time series of transactions on the basis of months for a total of 30 months. I wanted to test whether the fluctuations of the time series are random. I assume that the time series is stationary.

The test procedure I have in mind is the "runs test". It is a sequential test for randomness of alternative data as a non-parametric test. I would test every month, whether the time series increases (+) or decreases (-). Two examples of non-random sequences (from the literature).

[1] + + + + + -----
[2] + - + - + - + - + -

My question would be how a sequential test would be calculated using SPSS.


DATA LIST free / seq1 seq2.
BEGIN DATA.
1 1
1 0
1 1
1 0
1 1
0 0
0 1
0 0
0 1
0 0.
END DATA.
EXECUTE.

[Andy W]

This is available in NPAR TESTS.

DATA LIST free / seq1 seq2.
BEGIN DATA.
1 1
1 0
1 1
1 0
1 1
0 0
0 1
0 0
0 1
0 0.
END DATA.
EXECUTE.

NPAR TESTS
  /RUNS(.5)=seq1 seq2
  /MISSING ANALYSIS.

[Jon K Peck]

Analyze > Nonparametric Tests > Legacy Dialogs > Runs (NPAR TESTS)
or
Analyze > Nonparametric Tests > One-Sample Nonparametric Tests Test sequence for randomness

You might also want to look at Analzye > Forecasting > Autocorrelations


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        drfg2008 <[hidden email]>
To:        [hidden email],
Date:        08/22/2013 01:56 AM
Subject:        [SPSSX-L] Randomness of time series
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

I have a time series of transactions on the basis of months for a total of
30 months. I wanted to test whether the fluctuations of the time series are
random. I assume that the time series is stationary.

The test procedure I have in mind is the "runs test". It is a sequential
test for randomness of alternative data as a non-parametric test. I would
test every month, whether the time series increases (+) or decreases (-).
Two examples of non-random sequences (from the literature).

[1] + + + + + -----
[2] + - + - + - + - + -

My question would be how a sequential test would be calculated using SPSS.


DATA LIST free / seq1 seq2.
BEGIN DATA.
1 1
1 0
1 1
1 0
1 1
0 0
0 1
0 0
0 1
0 0.
END DATA.
EXECUTE.




-----
Dr. Frank Gaeth
FU-Berlin

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Randomness-of-time-series-tp5721672.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[drfg2008]

SPSS 20

Thanks!

However, it is a bit strange: One command shows no result.

This works:

NPAR TESTS
  /RUNS(.5)=seq1 seq2
  /MISSING ANALYSIS.


This does not give results (also no errors. Limited license?):


*Nonparametric Tests: One Sample.
NPTESTS
  /ONESAMPLE TEST (seq2 seq1) BINOMIAL(TESTVALUE=0.5 SUCCESSCATEGORICAL=FIRST
    SUCCESSCONTINUOUS=CUTPOINT(MIDPOINT)) RUNS(GROUPCATEGORICAL=SAMPLE
    GROUPCONTINUOUS=CUTPOINT(SAMPLEMEDIAN))
  /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
  /CRITERIA ALPHA=0.05 CILEVEL=95.

[Bruce Weaver]

NPTESTS produces that new-fangled output that has to be double-clicked, IIRC.  (I can remember the proper name for it just now--model viewer, or something like that?)

drfg2008 wrote

SPSS 20

Thanks!

However, it is a bit strange: One command shows no result.

This works:

NPAR TESTS
  /RUNS(.5)=seq1 seq2
  /MISSING ANALYSIS.


This does not give results (also no errors. Limited license?):


*Nonparametric Tests: One Sample.
NPTESTS
  /ONESAMPLE TEST (seq2 seq1) BINOMIAL(TESTVALUE=0.5 SUCCESSCATEGORICAL=FIRST
    SUCCESSCONTINUOUS=CUTPOINT(MIDPOINT)) RUNS(GROUPCATEGORICAL=SAMPLE
    GROUPCONTINUOUS=CUTPOINT(SAMPLEMEDIAN))
  /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
  /CRITERIA ALPHA=0.05 CILEVEL=95.

[Jon K Peck]

Both of these run fine in V20 for me.  Both of these commands are in the Base package, so there is no special licensing issue.  With NPTESTS, you need to double click the output object to see the full output, but the summary that you see in the Viewer has the test outcomes.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        drfg2008 <[hidden email]>
To:        [hidden email],
Date:        08/22/2013 07:21 AM
Subject:        Re: [SPSSX-L] Randomness of time series
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

SPSS 20

Thanks!

However, it is a bit strange: One command shows no result.

This works:

NPAR TESTS
 /RUNS(.5)=seq1 seq2
 /MISSING ANALYSIS.


This does not give results (also no errors. Limited license?):


*Nonparametric Tests: One Sample.
NPTESTS
 /ONESAMPLE TEST (seq2 seq1) BINOMIAL(TESTVALUE=0.5
SUCCESSCATEGORICAL=FIRST
   SUCCESSCONTINUOUS=CUTPOINT(MIDPOINT)) RUNS(GROUPCATEGORICAL=SAMPLE
   GROUPCONTINUOUS=CUTPOINT(SAMPLEMEDIAN))
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.




-----
Dr. Frank Gaeth
FU-Berlin

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Randomness-of-time-series-tp5721672p5721677.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[drfg2008]

Thanks for the hint.