Intelligence & Performance Management
Data Mining
If your business or organisation has been in operation for an extended period of time, you will have accumulated volumes of information about your business and your customers. Sadly, much of this information may be in old systems, inaccessible and of little practical use to you.
The information you are getting from your current database may also be of little value in its existing form. We can help turn your otherwise useless data into helpful information and knowledge about your customers and business processes.
Formal Data Mining initiatives can help you better understand patterns in consumer behaviour, review the success of past marketing strategies, and analyse transaction processing to help you to cut operational costs and improve organisational performance.
CIBIS also has Cognos Business Intelligence software to help with the analysis, report building and sharing of information. This makes our application tools and techniques amongst the most advanced available in the world today.
As a trusted and respected technology partner to many corporate and government clients we unlock your data and open your enterprise to exciting new opportunities. Much of what you will predict for the future is based on what you know about the past. To successfully plan for and manage your growth, you need to make the most of the historical information available to you.
Data Mining: A Case Study
CIBIS was recruited by a valued client in the higher education sector to data mine its enquiry, application databases and website logs, so that the organisation could better target prospective students and streamline the process of recruiting them.
Data Mining is exploratory in nature and its broad objective is to identify valid, novel, potentially useful and understandable correlations and patterns within existing data. The outcomes of the data mining process are geared towards finding new knowledge and developing greater understanding of the data, leading to improved decision making.
The steps below outline those usually followed in data mining activities:
-
Clean and pre-process the data: Different data sources use different conventions for data storage. We need to handle missing data fields, noise in the data, accounting for time series and known changes.
-
Integration of data sources: We have different sources of data and much common data between sources. The integration allows for easier modelling.
-
Selection: Select the relevant data which will be analysed.
-
Data Transformation: Choosing the data classification types. Narrowing the variables, applying dimensionality reduction and transformation techniques to find representations of the data that are invariant.
-
Data Mining: Choose the data mining algorithms, search for patterns of interest.
-
Pattern Evaluation: Interpretation of the patterns, perhaps hypothesis testing.
-
Knowledge presentation and reports: Consolidation of the knowledge discovered and report preparation.
We identified at least four basic sources of data from which we could develop a data warehouse.
-
External Agents: Call centre logs of student enquiries held significant information about a wide range of interactions with mainly prospective students. Phone logs, emails, messages, e-brochures, comments, referrals etc.
-
Information Systems: Existing systems holding information about already "converted" students. This data could be conveniently sourced from internal reports.
-
Web Site Logs: This is recorded knowledge including content from tracking tools, logs of served content, credit calculators and forms etc.
-
Subject Matter Experts: institutional information held by those working with the system; trends, practices and common processes.
Our objective was to better understand how and why students chose to study at this institution and what could be done to impact on efficiency. We determined that with the appropriate information our client could:
- Identify barriers in applying to the institutions
- Identify un-serviced market segments
- Better target their marketing spend
- Fine tune their knowledge of their existing database
- Increase their understanding of the market
- Refine the application process and associated recruitments tools
- Better manage the performance of their agents
- More accurately determine course quotas (predictive analysis)
- More accurately forecast revenues
Our Report
After rigorously attending to the processes outlined earlier, we were able to make a range of recommendations to our client and despite the technical limitations, we were able to discover some significant linkages between web site usage and ultimate application for study. Specifically, we were able to identify those pages and functions that were more likely to lead to new enquiries or applications being lodged. This lead to dramatic improvements in our client's understanding of the behaviour of web site visitors in relation to their decisions relating to studying at this particular institution.
