Skip to content

HoltWinters() - p-values for alpha, beta and gamma

3 messages · Kelvin Lam, Ken Lee, Ott Toomet

#
Need your expertise for the theoretical approach to
deduce the p-values for the level, trend and
seasonality parameters.  I wonder if there's source
code available.  Thanks group.

Kel
#
Dear all,
       Are there any problems at my BATCH program for the two error log, 
can I change the size of vector to avoid the error ?

error log 1:
Garbage collection 2881 = 2513+248+120 (level 2) ... 
1034265 cons cells free (35%)
14.1 Mbytes of heap free (33%)
Garbage collection 2940 = 2564+254+122 (level 2) ... 
1009265 cons cells free (34%)
15.4 Mbytes of heap free (35%)
Garbage collection 3006 = 2624+258+124 (level 2) ... 
984057 cons cells free (33%)
15.0 Mbytes of heap free (34%)
Garbage collection 3067 = 2678+263+126 (level 2) ... 
958953 cons cells free (32%)
14.6 Mbytes of heap free (33%)
Error: cannot allocate vector of size 2048 Kb
Execution halted

error log2:
Garbage collection 2895 = 2525+249+121 (level 2) ... 
1009160 cons cells free (34%)
15.2 Mbytes of heap free (34%)
Garbage collection 2962 = 2586+253+123 (level 2) ... 
983952 cons cells free (33%)
14.7 Mbytes of heap free (33%)
Garbage collection 3022 = 2638+259+125 (level 2) ... 
958848 cons cells free (32%)
16.1 Mbytes of heap free (35%)
Error: cannot allocate vector of size 448 Kb
Execution halted
#
Hello,

It looks like your memory has become fragmented (there are several MB
free but not a single 2 MB chunk. 

If this is the case, you may consider:

1) save your data, exit R, start it again and continue.  It should
   free all the memory, occupied by R, and hopefully allocate it
   better next time.

2) Use unix/linux.  There are rumours that unix has better memory
   management than windows (your mail was sent from windows).  

3) Dont start other programs while R is running.  Keep the number of
   running programs/precesses to minimum.  They may be part of the
   reason of memory fragmentation.

4) Try with subsets of your data, split your data into separate
   indipendent subsets (if you can) etc...

Perhaps it helps.

Ott

 | From: "Ken Lee" <ken_lee at tynesys.com>
 | Date: Fri, 18 Apr 2003 14:39:01 +0800
 | 
 | Dear all,
 |        Are there any problems at my BATCH program for the two error log, 
 | can I change the size of vector to avoid the error ?
 | 
 | error log 1:
 | Garbage collection 2881 = 2513+248+120 (level 2) ... 
 | 1034265 cons cells free (35%)
 | 14.1 Mbytes of heap free (33%)
 | Garbage collection 2940 = 2564+254+122 (level 2) ... 
 | 1009265 cons cells free (34%)
 | 15.4 Mbytes of heap free (35%)
 | Garbage collection 3006 = 2624+258+124 (level 2) ... 
 | 984057 cons cells free (33%)
 | 15.0 Mbytes of heap free (34%)
 | Garbage collection 3067 = 2678+263+126 (level 2) ... 
 | 958953 cons cells free (32%)
 | 14.6 Mbytes of heap free (33%)
 | Error: cannot allocate vector of size 2048 Kb
 | Execution halted