IS480 Team wiki: 2012T1 6-bit PD Business Intelligence and Analytics
by offering a place for exchanging ideas and information on its public domain.
|Home||Technical Overview||Project Deliverables||Project Management||Learning Outcomes|
|Technical Diagrams||User Interface Prototype||Business Intelligence and Analytics|
- 1 Chapalang Benchmark
- 2 Chapalang Analytics
- 2.1 Purpose
- 2.2 Operational Data Source
- 2.3 Analytical Techniques
- 2.4 Analysis of User’s Facebook Persona
- 2.5 Model for Personalized Dashboard
- 2.6 Outcome
CodeIgniter has a Benchmarking class that is always active, enabling the time difference between any two marked points to be calculated. In addition, the benchmark is always started the moment the framework is invoked, and ended by the output class right before sending the final view to the browser, enabling a very accurate timing of the entire system execution to be shown.
Chapalang Benchmark is a custom tool developed with the foundation of CodeIgniter’s Benchmarking class. Additional customization has been made to reflect extended measurement of execution for process-based series of activities, integrated with a database table to support other analytical execution.
One of the purpose of Chapalang Benchmark is to support an audit trail on the activities performed by users on Chapalang!. The tool collects the following set of data:
- Member ID
- Controller method
- Start timestamp
- End timestamp
- Elapsed time
By having Chapalang Benchmark to support an audit trail, we are able to trace the series of activities performed by a specific user on our system, given any time period.
While it may not be friendly to the privacy of users, and not efficient to design models customized to a specific user, it is useful to use the activity data collected to support advance analytics. At this stage, the data acts as a complementary operational data source for our analytics behind the Personalized Dashboard.
There are other usefulness of the activity data to discover system anomalies, user experiences, and more. Some examples are further elaborated below.
The purpose of Chapalang Analytics is primarily to support the low-level function, Personalized Dashboard of the portal. While that is the only application of Chapalang Analytics at this stage, it can be further expanded into other data-driven and user-oriented features which include but not limited to, target advertising and content filtering.
Operational Data Source
The analytics will make use of the following data:
- Facebook statuses, likes, post times, friends, gender, birthday, relationship status
- Chapalang Benchmark, which collects data on all user activities on Chapalang!, such as browsing history and past purchases
- Sentimental database, containing a list of words that characterises user’s personality and behaviour
The analytics will make use of the following techniques:
- Semantic Analytics
- Sentimental Analytics
- Rule-based Analytics
Analysis of User’s Facebook Persona
The first analytical step of Chapalang Analytics is to analyse a user’s Facebook persona. Through the analysis of the user-specific ODS, we will identify certain characteristics of a user.
Discoveries from Chapalang Benchmark
As we analyse the data from Chapalang Benchmark, there are some interesting discoveries that would otherwise be hidden or unknown to us.
Inconsistent elapsed time for view_product page
Despite using the same set of application files, the elapsed time for different products is different. We have some product pages which take as long as varying 15-30 seconds to load, while the rest are mostly under our acceptable response time of 3 seconds. Though we have yet to isolate the cause and find a permanent solution, the discovery will help us identify problems like these which may not be reported by users lead to a reduction in returning visitors.
Electronic products are most interesting to visitors
As we run more analysis, we realized that electronic products are most interesting to visitors because 87% of our visitors will view at 2 or more electronics products. They also spend longer time on electronics product pages than other pages on the portal. This might be useful for the company to offer more electronic goods deals, promotions or marketing.
Most traffic comes early in the morning and evening
From Chapalang Benchmark data, we also noticed a trend that most visitors, both registered and unregistered, visit our portal between 9am and 10am, and 3pm to 4pm. We are speculating that this might be because we have a user base consisting of SMU students and young working adults. The time window matches approximately the start time of classes in the morning and evening sessions. Also, it appears to be the first hour of work and evening tea-break time for working adults. Both of the reasons indirectly supports that they might be visiting our portal during break time and hence the company can hold hourly sales during these periods.
Model for Personalized Dashboard