Statisticus, Why and How?
Part II - Measurement Tools
by Hakan Eskici
2. Measurement Tools
There are simple ways to implement measurement tools into our web sites. The most
well known is the page counters which existed since the beginning of the web.
Web server logs are the second most well known tools of measurement, which are
stored for entire site and harder to analyse compared to the file access logs
which are generated by the tool that we are going to develop.
So, let's examine these measurement methods, and the ways to implement them
into our pages.
2.a. Page Counters
A page counter is the tool that you count the number of views of your page. Or is it? Let's put
it in this way: Should the page counters only count?
The answer is No.
Say, you developed a game site, and launched it one month ago. Your index page has 1,000,000
hits so far. So what? How could you tell how many hits are you going to get tomorrow? Or could
you tell the acceleration is negative or positive? If you only have a page counter, you cannot.
Here is a piece of code for a simple page counter:
<%
(...)
'Let's say we have different numbers for each page
lPageID = 34
'Open the connection
conn.Open sConnenctionString
'Execute increment
conn.Execute "UPDATE Counter SET PageCount = PageCount + 1 WHERE PageID = " & lPageID
'Close the connection
conn.Close
(...)
%>
This example assumes that we have a database table called Counter with two fields PageID and
PageCount, both of which are numeric data types.
2.b. Web Server Logs
Almost all web servers log the page requests along with some more data such as request date, time,
remote ip etc unless stated not to. These logs run for entire site and can be analysed by 3rd party
products. However, since we are trying to develop our own and a far better solution, let's forget
them for now.
2.c. Our Own Counter
The more data we have, the more information we can get.
Here is a quick seperation among data and information: Data is something raw, something not
processed, which cannot be visualized by human brain. Information is the valuable part of
the raw data we get by processing and analysing it.
However, in a world of information, we still can't risk the amount of disk space to store
large quantities of data. Moreover, we can't risk the processing time of the cpu's.
Fortunately, there is always a balance. By various optimizations, we can get somewhere between.
Now, we will define the pieces of information we want to store in the database.
But at first, we need to describe what we want to learn using this data.
Let's begin with simplest and go to hardest;
PageViews
» Total pageviews for entire site,
» Pageviews for each page,
» Pageviews for a specific time period (daily, weekly, etc.).
Visitors
» Total number of unique visitors for entire site,
» Visitors for each page,
» Visitors for a specific time period,
» Visitors for each page for a specific time period.
Averages
» Average Pageviews per Session.
» Average Stay Period per Session,
» Average Stay Period per Page,
In the third part of the article, we will get
into the details of our design and implementation.
Part I - Introduction
Part II - Measurement Tools
Part III - Simple Statistics
Part IV - Adding More Fields
Part V - Data Normalization
Part VI - Data Gets Valuable
|
English version
1544 membros
