[R-SIG-Finance] Approach to predict balance of account in retail bank

[Biswarup Ghosh]

Hi John ,

The data set I have is for one year . For some accounts which are very
active the balance is changed pretty often ,and for some it does not . Now
there is no apparent seasonality as this are bank accounts of retail
banking customers ( normal folks ) .
Now what would be your suggestion in this case .

On Sun, Feb 7, 2016 at 12:49 PM, John Frain <frainj at tcd.ie> wrote:

>
> On 7 Feb 2016 05:03, "Biswarup Ghosh" <reachbiswarup at gmail.com> wrote:
> >
> > Hi Daniel ,
> >
> > My confusion for using an Arima / or any other Time series  model, is,
> > since for every account I have a series  ,How do I ensure a model which
> is
> > applicable to all .
> > pardon my lack of knowledge in this.
> >
> > On Sat, Feb 6, 2016 at 5:28 PM, Daniel Melendez <
> > danielmelendez at alum.northwestern.edu> wrote:
>
> I would think that each individual account holder is different.  If you do
> find an appropriate model you will need to estimate separate parameters for
> each account holder. For each account holder I presume that the balance
> remains constant between the dates on which the balances are recorded. You
> should extend your data set by filling in the missing values by repeating
> the previous balance.
>
> Would it be sufficient to aggregate accounts for each date. You might find
> it easier to obtain aggregates (e. g. end of month aggregate data) for a
> period of several years and apply Bureau X12 methodology. I presume that
> data are seasonal.lpl
>
> '
>
> >
> > > Hello Biswarup -
> > >
> > > Have you looked into state space models?  If not, I'd first start with
> an
> > > ARIMA model.  I suggest working with the forecast package.
> > >
> > > Regards - Daniel
> > >
> > > > On Feb 6, 2016, at 6:23 PM, Biswarup Ghosh <reachbiswarup at gmail.com>
> > > wrote:
> > > >
> > > > Hi
> > > > I have a database table consisting of balance of accounts over a year
> > > > during different Business dates . I want to predict the balance of
> the
> > > > account in next one month period .
> > > >
> > > > Few of the concerns that I am facing is
> > > >
> > > > 1. The balances of this accounts are not on same dates like for
> example
> > > > some accounts is very active and the balance is changed over every
> one or
> > > > two days in this One year period and Some accounts are not that
> active
> > > and
> > > > rarely changes
> > > >
> > > > Now What I have done so far is
> > > > 1. Flagging no of changes of a account per year
> > > > 2. Percent of balance change on each period
> > > >
> > > > Now I am at loss on how to go with the forecasting algorithm.
> > > >
> > > > Now based on this parameters how I can give a forecasting , please
> point
> > > me
> > > > to a similar literature on which I can work on
> > > >
> > > >
> > > >
> > > > --
> > > > with warm regards
> > > > Biswarup Ghosh
> > > >
> > > >    [[alternative HTML version deleted]]
> > > >
> > > > _______________________________________________
> > > > R-SIG-Finance at r-project.org mailing list
> > > > https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> > > > -- Subscriber-posting only. If you want to post, subscribe first.
> > > > -- Also note that this is not the r-help list where general R
> questions
> > > should go.
> > >
> >
> >
> >
> > --
> > with warm regards
> > Biswarup Ghosh
> >
> >         [[alternative HTML version deleted]]
> >
> > _______________________________________________
> > R-SIG-Finance at r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> > -- Subscriber-posting only. If you want to post, subscribe first.
> > -- Also note that this is not the r-help list where general R questions
> should go.
>



-- 
with warm regards
Biswarup Ghosh

	[[alternative HTML version deleted]]