computing montly averages when time series observations are irregural

2 次查看(过去 30 天)
Dear all,
I have the following problem
Let
A ={
'02/11/08' [0.2150] [429.5674] [ 100] [2.2678e+04]
'30/11/08' [0.2117] [463.1360] [ 100] [2.4383e+04]
'28/12/08' [0.2209] [436.6316] [ 100] [2.3148e+04]
'25/01/09' [0.2212] [441.5430] [ 100] [2.2987e+04]
'22/02/09' [0.2201] [453.7015] [ 100] [2.3675e+04]
'22/03/09' [0.2185] [461.3925] [ 100] [2.3680e+04]
'19/04/09' [0.2104] [486.4095] [ 100] [2.5367e+04]
'17/05/09' [0.2162] [451.4833] [ 100] [2.3986e+04]
'14/06/09' [0.2158] [475.8620] [ 100] [2.4245e+04]
'12/07/09' [0.2211] [449.4574] [ 100] [2.2766e+04]
'09/08/09' [0.2221] [456.2507] [ 100] [2.2523e+04]
'06/09/09' [0.2175] [472.2659] [ 100] [2.3593e+04]
'04/10/09' [0.2182] [479.3408] [ 100] [2.4359e+04]
'01/11/09' [0.2286] [442.6719] [ 100] [2.1490e+04]
'29/11/09' [0.2211] [481.4548] [ 100] [2.4054e+04]
'27/12/09' [0.2259] [468.2757] [ 100] [2.3037e+04]
'31/01/10' [0.2300] [461.5581] [ 100] [2.2050e+04]
'28/02/10' [0.2259] [487.6257] [ 100] [2.3293e+04]
'28/03/10' [0.2200] [529.8777] [ 100] [2.5493e+04]
'25/04/10' [0.2315] [433.3039] [ 100] [2.0387e+04]
'23/05/10' [0.2274] [500.8603] [ 100] [2.4019e+04]
}
As you can see from the first column I have irregular time series observations. Each observation can represent a 4-week or a 5-week or a 6-week average. For example the observation [0.2117] which is the second value of the second column is the average of the values that have been recorded between ‘03/11/08' and '30/11/08' In order to be able to analyze econometrically these data and run my regressions I need to obtain regular time series observations. So I need to convert these values to monthly averages and have something like
Let A ={
'11/08' [mon. average] [mon. average] [mon. average]
'12/08' [mon. average] [mon. average] [mon. average]
'01/09' [mon. average] [mon. average] [mon. average]
'02/09' [mon. average] [mon. average] [mon. average]
'03/09' [mon. average] [mon. average] [mon. average]
'04/09' [mon. average] [mon. average] [mon. average]
'05/09' [mon. average] [mon. average] [mon. average]
'06/09' [mon. average] [mon. average] [mon. average]
'07/09' [mon. average] [mon. average] [mon. average]
'08/09' [mon. average] [mon. average] [mon. average]
'09/09' [mon. average] [mon. average] [mon. average]
'10/09' [mon. average] [mon. average] [mon. average]
'11/09' [mon. average] [mon. average] [mon. average]
'12/09' [mon. average] [mon. average] [mon. average]
}
Could interpolation be a solution to this problem? I would be grateful if you could provide some code.
cheers
  3 个评论
antonet
antonet 2012-7-29
编辑:antonet 2012-7-29
Hi imager! yes as you can see there is overlapping across the dates as each value is the average of the values that have been observed 4 weeks (or 5 weeks or 6 weeks) back. Advice me what is the best approach to calculate the monthly averages. PLease ask me if you require further information

请先登录,再进行评论。

采纳的回答

per isakson
per isakson 2012-7-28
编辑:per isakson 2012-7-28
Hint:
  • vec = datevec( A{:,1}, 'dd/mm/yy' )
  • out = accumarray( vec(:,2), A(something), [], @mean )
Better
vec = datevec( char(A{:,1}), 'dd/mm/yy' );
out = accumarray( vec(:,2:3), transpose([A{:,2}]), [], @mean );
This returns the means of the values of each month respectively. There is a value for each month. Is that what you want?

更多回答(0 个)

类别

Help CenterFile Exchange 中查找有关 Calendar 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by