Sample Beginner Analysis Command Strings

For discussion of how to use the Cycles Research Institute' own Cycles Analysis & Time Series software.

Sample Beginner Analysis Command Strings

Postby Blair » 23 Aug 2010, 02:42

Here is a simple set of commands to get people started with the use of CATS.

For this example, my file is number 22, and it is daily data.
Also, I am using the Hurst Nominal values of 20, 40, and 80 days for the Cycles to be used in the Regression analysis.

Description: Load the data file, Copy it, Generate a Moving Average of 162 days, Centre the Moving Average data by 81 days, Name the Centred Moving Average, Look at the stack Elements
Commands: f22 l @162 :-81 N"CMA" ?

Description: To Look at how your CMA compared to your data, just Graph them
Command: g2

Description: Perform a Spectral Analysis on periods Greater Than 20 ***** Must view the speclog.csv file (in your currnet data directory) after running the Spectral Analysis to see the results
Commands: S20

When Viewing the speclog.csv file, I sort on the 5th Colum (last) which is the Bartell's value. The first two elements have always been "assumed" cycles, so I ignore them. Then I try to estimate what reasonable values are to use from the grouping of the values in the second column (Number of periods in the cycle - since I am using daily data - it is number of days in the cycle).
Again, for this example, I will do regression on the 20, 40, 80 day cycles according to Hurst Nominal Model. In a real analysis, you will need to substitute your values for the 20/40/80 I have used, as well as deciding what is the appropriate Averaging period.

NOTE commands are CASE SENSITIVE. Capital or Upper Case letters and lower case letters are specific. And yes, I found out the hard way.... :oops:

Description: Delete the spectral Analysis (top element of stack), create the Sine and Cosine elements for all 3 cycles, Load the original source data, perform the Regression, perform the Matrix multiplication, Name the Result
Commands: z-1 20 Usc 40 Usc 80 Usc l1 R6 M6 N"RegMult"

Since the Regression and Multiplication is over the Entire Data Spectrum, not just your data elements (f22 is from 19950314 to 20100813) you need to set the valid dates for the RegMult. The Start Date needs to be the same, but the End Date can be later than your data runs - hopefully to enable a projection - (But I have not figured out yet how to do that). Then take a look at the result (hint it is just a few cycles).
Description: Set the valid date range and look at the results
Commands: D19950314,20200101 g

Description: Locate or estimate the midpoint of the graph. Duplicate the Regression, Subtract this value from the Regression, and then add it to the CMA, and then graph the summed value against the original values.
Commands: l 1254.2 - l2 + N"Summed" l1 g2

You should have a plot of your original data, and a Summed Centred Moving Average with the 3 cycles Regression added in. Hopefully you will have reasonable results.

NOW, just need to figure out how to Project these into the future.

So, here is the summary of commands. Extraneous "?", "g", "z" commands have been removed. I assume if you could type it all in as one line when you get good. I still need to N"name" my stack elements and "?" see them, and "g"raph them.

f22 l @162 :-81 N"CMA" S20 z-1 20 Usc 40 Usc 80 Usc l1 R6 M6 N"RegMult" D19950314,20200101 l 1254.2 - l2 + N"Summed" l1 g2

NOTE: the lowercase "l" and the number "1" can look quite similar.

I hope this summary helps get more people working with CATS.
Blair
 
Posts: 11
Joined: 05 Aug 2010, 14:25

Re: Sample Beginner Analysis Command Strings

Postby Martin » 15 Sep 2010, 12:25

Nice run through, confusing but nice!

You said, "Description: Locate or estimate the midpoint of the graph. Duplicate the Regression, Subtract this value from the Regression, and then add it to the CMA, and then graph the summed value against the original values.
Commands: l 1254.2 - l2 + N"Summed" l1 g2"

What is the purpose in doing this step?

and yif you want to trim the dates what I do is, after entering the cycles with the Usc, then check the stack and set the date of the top line to a specific start date and then run it into the future a little. Then I call up the original data like this; L1 dstartdate,enddate. The end date cannot go beyond the date you have data for. This date will be shown on the stack in L1. then I simply G2

Hope this helps.
Martin
 
Posts: 5
Joined: 15 Sep 2010, 12:15

Re: Sample Beginner Analysis Command Strings

Postby Blair » 15 Sep 2010, 13:13

Martin,

That step give me a midpoint value around which to plot the amplitude of the 3 summed cycles. Without it, the CMA would be based around the 1254.2 value, and the 3 cyles would be in the 1's and 2's area - essentially a flat line near the zero when plotting.

Still trying to figure out when the best time is to trim the dates....

Glad somebody found this posting...

Blair
Blair
 
Posts: 11
Joined: 05 Aug 2010, 14:25


Return to CATS Software

Who is online

Users browsing this forum: No registered users and 1 guest

cron