Thanks - I should have clarified that the x axis wouldn't show date but would
instead show days since the first purchase. I plan on calculating this as a
separate columnn and then using it to graph with. As you say, plotting by
date will result in a mess.
Is it possible to automate the process of generating all of this data
automatically? I don't think Excel will automatically create the charts I'm
looking for, so I would have to manually add each new customer series one by
one and then tweak the series formula. This is not viable from a sheer time
perspective.
The only way that comes to mind is write some sort of script to go through
the data row by row. As soon as the script detects a change in the customer
field, it will create a new series out of the rows it has previously
processed.
If this doesn't work then the only other option is to use a more flexible
graphing tool such as gnuplot or something in combination with scripts to
process and prepare the data for graphing. I would have much more
flexibility in processing and graphing the data but I was hoping that Excel
would be powerful enough to do this. It can certainly do a lot - sometimes I
think it can read my mind. :)
"Jon Peltier" wrote:
One of your buying patterns was when customers bought widgets relative to
their first widget purchase. If you just plot by date, you'll have a big
mess. But try subtracting initial purchase data from purchase date, so you
have number of days since first purchase. This will give you a better X
variable. For your Y variable, you may leave quantity as is, or you may
divide quantity by initial purchase quantity, or you may transform it in
another way. And don't treat this as either-or. Make several charts, one for
each variation you can think of. And keep trying other variations as they
pop into your head.
- Jon
-------
Jon Peltier, Microsoft Excel MVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. - http://PeltierTech.com
_______
"rparker" wrote in message
...
I was thinking about using an xy scatter plot because I wanted to plot time
on the x axis and number of widgets purchased on the y axis. So the x
axis
is definitely treated as a value rather than a category.
I actually don't care too much about differentiating among the 255
customers; I'm mainly looking for general trends in buying behavior. I
was
thinking that if I graph a large number of customers and juxtapose their
buying behavior together, certain patterns will emerge.
Sparklines is an interesting concept. The biggest issue I see with them
is
that they don't allow me to map many simultaneously together, which is
what
I'm primarily interested in. The aggregate is more important than the
individual customer.
"Jon Peltier" wrote:
How are you going to differentiate from among 255 customers on a chart?
What type of chart is it?
I wonder if using a set of sparklines might be what you need. These
people
have a good product:
http://www.bonavistasystems.com/Prod..._Overview.html
- Jon
-------
Jon Peltier, Microsoft Excel MVP
Tutorials and Custom Solutions
Peltier Technical Services, Inc. - http://PeltierTech.com
_______
"rparker" wrote in message
...
I want to track the widget buying history of each customer. I already
have
sales data on how many widgets have been purchased by a distinct
customer
on
a specific date. There are other helpful fields such as geographic
region,
customer size, etc. I've already been able to massage the data a bit
and
use
pivot tables so I can get statistics such as total widgets per region
per
quarter, widgets per quarter purchased by new vs. existing customers,
etc.
I'm trying to prove or disprove a hypothesis about the data; namely,
are
most widget purchases made simultaneously with the initial sale of the
first
widget or afterward? I was hoping that by graphing the widget buying
history
of a number of customers, this would become more apparent. This is why
I
planned to make each customer into its own series so I could graph out
the
customer's buying history over time.
I plan on segmenting the customers by geographic region and customer
size.
If I still have more than 255 customers per segment then I was simply
going
to make each segment smaller by tweaking the customer size categories.
I'm doing this analysis mostly for myself so I can draw some
conclusions
about our customers and their buying patterns, so you are correct in
saying
that legibility isn't the foremost concern.
Hopefully my questions make more sense! I'm still fairly new to Excel
charting but I'm doing my best to learn as much as I can to get the job
done.
"Jon Peltier" wrote:
It depends on what you want to show. As I asked...
Does each customer need its own series? Are you interested in
widgets per day, cumulative widgets, or something else?
Another thing you need is legibility, but your posts don't indicate
that
you're too concerned with that.
What you need to think about includes:
* do I understand my client?
- do I understand the data?
- do I understand what they need to learn from the chart?
- do I know how they currently display the data?
* what are my plotted values?
* how do I segment the data?
- what are the categories? dates? parts? customer? something else?
- how do I define my series? customer? part number? classifications
of
parts?
- Jon