From CS294-10 Visualization Fa08

Browser Statistics

Contents 1 Domain 2 Data Source 3 Preprocessing 4 Initial Question 4.1 Visualization 1 4.2 Visualization 2 5 Question 2 5.1 Visualization 3 6 Another classification

Domain

Everyday I am using a Web browser. But, I almost do not know what other people use a Web browser. Some Web browsers may be better than others, So I changed my browser manytimes. But, recently I did not change my browser, because it took a long time to be accustomed to new web browser. So I would like to other people's browser usage to motivate changing my browser. Therefore, I decided to focus on the transition of Web browser usage.

Data Source

To answer this question, I used statistics data as follows w3schools Browser statistics It provides browser statistics and other details about its visitors. (This site provides training on web standards as well as some proprietary technologies, so statistics may be biased.)

Preprocessing

The statistics is provided by html table, but each column label is inconsistent for dates. So I formatted all of this data to work with Spotfire. I rearranged the initial data table to consistent labeled column corresponding to dates that are used as databasekey. With this, I could easily import the data into Spotfire.

Initial Question

My initial question is how share of web browser has been going through for the years. Although it is very ordinal question, the answer leads me to interesting result and gives me another question.

To answer the question, first I used scatter plot to glance the distribution of data. In scatter plot I put Date to x-axis and all web browser usage to y-axis. As a result, I got the Initial visualization as follow.

Visualization 1

There seems to be some correlations here, but it is not clear enough to be certain. To make their correlations clear I thought a line graph would be better represent the data , Since the data displaying the trend of prevalence in browser usage are substantially continuous and time sequence value. Thus, I obtained the second result as follow.

Visualization 2

From this line chart, I can see there is two gap on it. One gap exists between Firefox and Mozilla Suite and another gap exists Mozilla Suite and Netscape5. I remembered users are continuing to use the same or similar series of web browser. So, I thought it would be better introduce a classification to the visualization and hit on second question.

Question 2

we know browsers are able to be placed in some categories ,for example series of Internet Explorler. If we aggregated the each browser usage in its classification, is there a correlation between categories?

To answer the question, first I grouped web browsers as as flows.

1. Series of Internet Explorer : Internet Explorer
2. Series of Mozilla and Netscape  : Firefox, Mozilla suite and Netscape
3. Series of Opera : Opera
4. Everything else

Moreover, I aggregated the each browser usage in its classification and labeled each summation. Finally emphasize the effect of grouping in visualization, I put the same color corresponding the categories, which yielded the following result.

Visualization 3

Another classification

Some people may think this classification is not good for representing taxonomy of browser. Actually, I think other classification such as layout engines(Trident, Gecko, WebKit etc). Since some web pages are too optimized for certain web browser, we can not see them right or sometimes can not use any functions such as java script ,flash, and so on. In fact, I tried the another classification as follows Category in layout engines

1. Trident-based : AOL. Netscape, Internet Explorer
2. Gecko-based : Mozilla suite, Firefox, Netscape
3. Presto-based : Opera
4. WebKit-based : Safari
5. Everything else

However, the result looks very meaningless and there seems to be no correlations.

---