analysis of web logs to understand how users use the website

Understanding how users use the website is important for improving the ease of use.

Suppose I have 3 website users (1, 2, 3), three webpages ("A", "B", and "C"), and each user visits three pages.

In this post I outline different data structures available for capturing user activity.

A matrix of pages and page visit order

	page A	page B	page C
first page	1, 2		3
second page		3, 1	2
third page	1, 3	2

Here time moves from the top row towards the bottom row.

The steps taken by a single user are easy to trace.

A Markov model

	A	B	C
A	0
B		0
C			0

A list per user

user 1: [A, B, A]

user 2: [A, C, B]

user 3: [C, B, A]

Same information as present in the matrix. 

Here time moves left-to-right

The list length can vary per user.

List of tuples per user

As a modification to the list, each element could include the page name, the render time, and the dwell time:

user 1: [(A, 0.2, 55), (B, 0.3, 20), (A, 0.4, 126)]

user 2: [(A, 0.1, 65), (C, 0.2, 234), (B, 0.4, 23)]

user 3: [(C, 0.3, 15), (B, 0.1, 53), (A, 0.3, 45)]