The consumer market is adapting to the thought of using m-business in their daily life. The convenience and opportunities that data mining can provide to m-business are nearly unlimited. There are many areas, including customer relationship management (CRM) and marketing, where currently many commercial e-business application developments and researches are conducted actively (Bowi, 2002; Magic Software, 2002; MobileIN.com, 2002; Regisoft, 2002; Sarwar, Karypis, Konstan, & Riedl, 2000). The possibilities of mining the mobile data sector include nothing less than guiding a nascent industry to fulfil its promise. This section attempts to summarise some applications that data mining can assist in the growing field of m-business.
Geographic aspects like location and temporal aspects are very interesting in the mobile environment. The GPS (global positioning system) mobile technology has allowed for the identification of the location of the users at a time (Cousins & Varshney, 2001; Duri, Cole, Munson, & Christensen, 2001). By mining location information, we can see the subscribers' behaviour over time. For example, knowing the locations that a person frequents will allow the prediction of a user 's daily life. If a person is an office worker, it can be noted that most of the time in a week, this person should be present in the work area and at home. Then we can offer location-based services that meet their needs.
Since a phone company already knows some personal and demographic information for all subscribers. With knowing the location- related information, we can generate a dataset with attributes such as subscriber ID, age, gender, marital status, employment, income, place of location, time of location, etc. The clustering data mining technique can be used to group subscribers according to their similar interests and can predict what the personality of the person is. Therefore appropriate location-based services can be offered to specific groups of subscribers. For example, if a person is most often sighted in supermarkets and department stores and at home and is seen shuttling between sales events, then this person can be classified as a possible homemaker who is generally interested in events that offer a discounted purchase. Information about sales events near to their place of residence or visit can be sent to a group of subscribers or to an individual subscriber according to their classification. Another example is if a person is often present in the locations of educational institute, pubs, and concert events and the age group is between 20 to 30 years of age, the person can be inferred as an outgoing person. Likewise, information about music events can be sent to subscribers who have an interest in music, or information about nearby takeaways can be sent to office workers working in late hours. Later on it can be analysed whether subscribers are taking advantage of the information suggested by operators, so that operators can take decisions and actions to improve services or increase revenue.
Knowing the location and time of visits for each subscriber, associative data mining can also be used to indicate which places a person is most likely to visit in a single trip or in two consecutive trips. This information can be used to suggest a new person to visit the place B if the person is on the place A based on experience of previous visitors .
In terms of a business-to-consumer (B2C; Varshney, 2001) relation, such information will allow business to provide the appropriate marketing information to the specific category of users. For example, users that are categorized as the type most interested in the latest financials news are unlikely to be interested in the information like the latest gardening tools. Thus, a business is better able to identify the needs of users and customize its marketing services to the users. Providing all information to every user, ignoring the user's interest in the type of information given, can result in the user unsubscribing to the services offered by the business.
Issues pertaining to a business-to-business (B2B; Varshney, 2001) relation are also essential to consider. The ability to track the location of employees is ideal to a business to determine the work efficiency of employees. For example, if the worker is sent to a client's location to perform certain duties for the day, the worker's whereabouts are known. If the worker does not go to the client location as directed, that will show the possible work attitude of the worker. Thus, it is ideal to mine the analysis of the worker's work efficiency and attitude to determine the worker who is performing the best and is most suited for the rewards. With the ability to track the location of employees, we do not just have the input attributes, such as employee details, duties performed, and time to finish the duties, but we also have attributes like locations and associated times to perform the duties . We have such information about previous employees too. This previous information can be used as an input dataset in predictive data mining , which recognises the distinct characteristics within the dataset due to which a worker is supposed to be working efficiently . Based on this information, a new employee's performance can be measured and suggestions to improve the work efficiency can be provided.
Businesses like courier companies are highly dependent on information regarding the locations of the transported parcels. Having the knowledge of the locations of the parcels allows courier companies to make more informed decisions. For example, if a parcel is to travel between locations A and B, the courier company can do a comparison between the same situation over the last few months using predictive data mining and determine if the operation has been efficient or not. If the time to transport a parcel between two locations is much longer for a particular situation, the management can enquire whether it is due to a new staff at a checkpoint that is slowing down the operations. Or is it due to the driver making mistakes, travelling a much further route? Such timecritical issues are vital to a courier company to keep up the standards of their services, which determine the future survival of the company.
Managing information is becoming one of the biggest issues today. It is important to provide the information that the customer is interested in, referring to the personal life situation and lifestyle. Personalization is particularly useful in areas like information sourcing, whereby data mining can be used to source information that the user specified. Short message service (SMS) is used primarily for simple person-to-person messaging. An increasing number of mobile information services, such as news, stock prices, weather, and notification of e-mail and voice, enables one to aggregate SMS technology and DM technology for providing as one-to-one or one-to-few information services. Information, obtained from analysing the user data about accessing these services using predictive and clustering data mining , can be used to create personalized advertisements to the customers, delivered by SMS.
Data mining has already been used in e-business to personalize or customize information tailored to the needs of users. M-service providers can also use personalization to provide the products or services that match the needs of individual users. For example, the service provider can use a clustering data mining tool to select hotels, either five-star hotels or three-star hotels or hotels in the suburbs or downtown according to the customer preference, based on previous similar travel service experiences. The clustering data mining technique groups customers with similar preferences. When a new customer mentions his/ her preferences, a similar preferences cluster is matched, and based on those preferences a recommendation is made.
Relevant services can be offered based not only on the personal profile of the device holder but also on the device holder's location and time factor. For example, m-business applications used in the travel industry can assist users to find attractions, hotels, and restaurants of their preference on requested location and time. Associative data mining can be used to indicate which places a person is most likely to visit in a single trip or in two consecutive trips, with having inputs such as location and time of visits to attractions for each user. This provides great convenience for users as these services can be used while driving; for example, a suggestion can be given based on the association rule that if the user is on place A then the user should visit place B since 80% of previous visitors have done so.
Due to the limited screen space provided on mobile devices, it is difficult for mobile users to browse the product or service catalogues on the devices. Thus, not only personalization services offered by the vendors can provide great convenience for customers, these services can also assist to retain customers that are critical in today's competitive environment.
Any m-business providing mobile facilities constantly analyses whether the profits derived from the business are sustainable. Data mining is essential to do a trend analysis of the business over a period, for example, to determine if the progress of the business is of a reasonable expectation. One of the techniques to determine the profitability of the business is the calculation of the return of investment (ROI). ROI is typically calculated as: (average returns / average investment) * 100% (Dickerson, Campsey, & Brigham, 1995). A possible way to analyse the data collected from the profit derived from the business is the use of linear regression (a value prediction technique of the DM) to predict the prospect of the business. A graph can be plotted to provide visualization and to allow the analysis of the future and potential of the business. Figure 2 illustrates the graph based on average returns verses average investment into the business. Many other business factors can also be considered during regression analysis.
Scenario 1, the best-fitting line derived from a variation of investments appears as Line A. This shows that the investments made into the business have been worth undertaking, as the increase in profit is greater than the increase in investments into the business.
Scenario 2, the best-fitting line appears as Line B. This means that the average return derived is equivalent to the average investment. Thus, it is neither profitable nor detrimental to carry on the business.
Scenario 3, the best-fit line appears as Line C. This means that the cost incurred to provide the mobile marketing service is greater than the revenue generated from the business. This shows that the mobile service provider has to rethink its operational and management plans.
In the near future, the PDA or the mobile phone will be our wallet, as most services will be payable with a few clicks on our devices. The service providers will have an enormous amount of consumer behaviour information stored in their databases. Patterns in consumer behaviour can be analysed from many angles, and information can be extracted to benefit further business implementations . M-business with DM provides new channels to marketing that can be used to provide a much more direct and to the point advertising to the customers.
Service providers can analyse the data (e.g., by analysing gateway log files and content server log files on WAP) and predict the consumers' buying and usage patterns or understand how mobile subscribers used their wireless services. Using this data, companies can apply data mining to identify customer segments using clustering data mining techniques, to distinguish customers' consumption patterns using deviation detection techniques , and to predict transaction trends using associative data mining techniques . This information then can be used to provide better services to the customers or to attract potential customers.
For example, if customers are buying several services, predictive data mining can help the service provider to forecast the users' service needs. Users seem to change their usage and needs; these changes need to be captured and analysed to predict what the users want/need in the future. Inputs to the predictive data mining process will be the details of users (personal, demographic, and geographical) who are using the services and also the details of users who are not using the services. The predictive data mining engine will be able to establish rules, based on distinct features, why certain users are subscribing these services.
If customers are using a combination of two or more services, associative data mining can help the company to provide better services to subscribers. Providing inputs such as customerID and the services used by them, associative data mining can infer rules such as 75% ofcustomers who use services A and B also use services C and D. This will help service operators to adjust prices on packages to get more customers to use all. These are just a few examples. Essentially, the customer purchasing information is stored on a digital storage medium and this gives the m-businesses new opportunities to increase their markets.
All the patterns between mobile phone brands and usage of available services can be extracted and used to further marketing. Data to analyse in this situation are the amount of mobile phones in the market, how many of these users use the services available, and how much usage, measured in currency, that the average user uses. This data analysis using classification or regression data mining techniques will be able to predict the trends and patterns of usage of mobile phones and mobile services.
For example, the most popular services bought through m-commerce technology are mobile ringing tones, logos, and screensavers. The most commonly used interfaces for these kinds of transactions are short message service (SMS) and the standard e-commerce interface, the Internet. An example is Nokia's focus on screensavers, logos, and ringing tone availability. This is most likely to be a result of previous research on their users' trends, by capturing the data on users' demands and needs and then analysing users' feedback. This information helped Nokia to develop a new market product where the product is no longer just a mobile phone but also provides extra features like SMS, logos, and additional ringing tones and screensavers (Nokia, 2002). These new features can be bought using the phones. All these transactions between the users and the service providers, and the patterns of customers buying ringing tones, logos, screensavers, SMS pictures, etc. can be analysed for further actions by the use of data mining tools.
A mobile commerce platform should integrate with existing back-end databases and businesses applications to deliver data via all the channels such as WAP, VoxML, TruSync, Bluetooth, or any wireless protocol. Data mining can be used to match which channel is best at a time to deliver the information. Data mining can optimise the amount and format of the content for delivery based on the connection speed of the device requesting the information. Data mining helps to decide what tasks , activities, and transactions are most economical and beneficial to use.
Unlike e-commerce in which the customer accessing the company site is anonymous, it is easy to identify the customer doing m-business and track their behaviour from various sources to build their profiles. This information can be used for marketing purposes. Clustering data mining techniques can be used to segment groups of people that appeal to certain products or services. Knowing this information will allow the marketing department to come out with "gimmicks" to target specific customers. Also, associative data mining techniques can be used to identify certain items that customers tend to buy together. This information will allow m-businesses to offer discounts on various purchases without bearing any losses, in turn attracting more customers and revenue.
The effective and subsequent analysis of the types of fraudulent activities in telecommunication systems is one of the applications that data mining can assist in a mobile environment. The dynamic nature of different fraudulent activities and the changes of the normal usage can lead to detection of fraudulent activities through observing behavioural patterns. A data mining system will have plenty of examples of normal usage and some examples of fraud usage. Based on these previous examples, a predictive data mining system establishes facts about fraudulent activities. Whenever a change in the normal usage is detected , the system analyses the change and will be able to predict whether the change is a fraud or not.
Churn management is a term used in the telecommunication industry to describe the process of ensuring that profitable customers stay with a particular company. Predictive data mining techniques can assist in churn management by forecasting whether a given individual is likely to move to another service provider by analysing their usage patterns, and these techniques are able to define the correct actions to keep that profitable customer (Purba, 2002; SAS, 2002).