Ad hoc analytics
Examples of ad hoc analytics / Market Analysis / Forecasting and simulation, or "what if"analysis / Identifying key causal factors affecting an area of interest / Risk and Quality Management / Optimization / Identifying trends over time / Cluster and pattern analysis / Raising alerts about new issues requiring immediate attention / Database Marketing / Fraud Detection / Data Quality Management / SuperSTAR / The power of OLMAP put to use / Easy querying
Ad hoc analytics can range from simple data exploration to advanced analytic modelling and data mining. Analytics may be applied as appropriate to summary or detailed data, or a combination thereof, including direct application to very large volumes of detailed data.
Typical areas of usage are:
-
<Market analysis, including profiling of customers and others
<Forecasting, and simulation of "what if" analyses
<Identifying key causal factors affecting an area of interest
<Risk and quality management
<Optimization
<Identifying trends over time
<Cluster and pattern analysis, including rules of association
<Raising alerts about new issues requiring immediate attention
<Database marketing
<Fraud detection
<Data quality management.
Analytics is a large and very complex area of mathematics, with different mathematical and statistical techniques being appropriate to different data sets and different types of usage. If a technique is applied to an inappropriate data set, then misleading results will be produced.
Our Client Services consultants can work with you and your staff to identify and apply the appropriate techniques to meet your needs. This includes 'duty of care' to ensure that the techniques being applied are appropriate for your data, so that you can then rely upon the results being produced. Particular attention will be paid to ensuring that you and your staff can easily understand the results, and are not swamped with data.
Our Client Services consultants can also work with you and your staff to embed these analytics into your business workflows, minimizing the amount of manual work done in transferring data to and from your other systems. The interfaces to your process management and reporting systems can also be automated. This will enable you to improve staff productivity, while simultaneously improving the quality of the work done.
| Create a query to identify a customer segment of interest. |  |
Analyze this segment with a range of canned summary statistics. |
|
||
 |
|
 |
|
||
|
||
| Explore relationships using techniques such as ColourVIEW. | |
Export the records for use in sales campaigns or to investigate cases or incidents of fraud. |
|
||
 |
|
 |
SuperSTAR has the ability to create synthetic or derived variables on the fly. This enables 'what if' type modelling to be conducted easily. For example we may want to look at the effect of a 10% tax break for a certain demographic (Non-europeans). We want to know how many peoples net income will increase to over $50,000.
| We take the net income variable for each person and add 10% to it. | |
We then band the income to identify over $50,000. |
|
||
 |
|
 |
|
||
|
||
| We then create a tabulation to show how many Non-Europeans now have a net income of over $50,000. | |
We can then extract out the unit records of these 28 people. |
|
||
 |
|
 |
We can provide you with the analytics required to better understand the existing and potential markets for your existing and potential products, including analysis by geographic area, time, customer profile, product type, and market penetration.
We can help you answer questions like: "Which products are being purchased by which customers, and when? How is each customer's purchasing profile changing over time? Including after sales service, which customers are the most profitable, and the least profitable? What are the key factors impacting customer retention and profitability?"
Forecasting and simulation, or "what if" analyses
By applying a forecasting method that is appropriate to your data, we can help you answer questions like: "Based on current trends, what sales and profits can be expected for each product into the future? If we increased our marketing expenditure on product X by $Y, what would the likely sales volume and profit be? What would be the impact of discontinuing production of product Z?"
Identifying key casual factors affecting an area of interest
Causal analysis can be very powerful in helping you and your staff answer questions like: "What event patterns typically precede a customer leaving? What time period is typically available to resolve customer issues before they decide to leave? What operating conditions typically precede equipment failure? What are the signals that a piece of equipment is likely to fail soon?"
There are many other areas of application, for example, analysis of click streams on web sites to determine when a visitor is about to leave the site, and what information should be provided before this happens.
Our analytics can help you and your staff to track key risk and quality metrics against goals using management dashboards and reports, then drill down to detailed data so as to understand the root causes of problems.
You and your staff can also be automatically alerted to new issues arising when underlying data moves outside preset boundary values, for example, when the pressure in a pipeline rises above a preset value.
Optimization enables you to tune your business so as to produce the best possible results for you and your customers, enabling you to answer questions like: "Given the results of a test mailing to a random selection of customers, which customers should we now mail to ensure a maximum profit on the campaign?", "What is the best product and marketing mix for each geographical area?", "What is the best distribution network for our products?", "What is the optimal amount of shelf space for our products in retail outlets?", "Where is the best place to locate a new manufacturing plant?"
These trends may be associated with normal calendar time, or the time relative to a specific event, for example, the purchase of a product.
Typical questions asked are "What are the seasonal trends in market uptake, and what are the underlying non-seasonal trends?" "What are the key factors that are likely to be associated with product uptake over the next two years?" "When are the key danger periods for customer dissatisfaction after product purchase, and what are the signs of dissatisfaction?" "If we undertake maintenance procedure X, then how long is the piece of equipment likely to run without failure?"
Cluster and pattern analysis, including rules of association
This can be done visually by exploring the data, for example using ColourVIEW, or automatically by our analytics software.
Typical questions asked are "Which groups of customers have similar usage of the hot line, and what are the key characteristics that distinguish these groups?" "Which events frequently occur together?" "Which product purchases are likely to be followed by other product purchases?" "What usage of the web site is most likely to precede product purchase?"
Raising alerts about new issues requiring immediate attention
This can be a very powerful technique, enabling you and your staff to fix problems at their inception, before they have serious consequences.
In general terms, the analytics software monitors incoming external and internal data streams, and automatically raises an alert when a particular event occurs, for example, a fraudulent person contacts your organization, or when a specified data item moves outside preset thresholds, for example, when the operating temperature of a piece of equipment becomes too high.
Alerts can be applied can be applied generally, for example to monitor stock levels, or specifically, for example, identifying particular customers which are likely to leave if their issues are not rapidly addressed.
The database is analyzed, this in conjunction with external reference data as appropriate, to identify the key characteristics that determine which products are best marketed to which customers, and how.
Typical results are demographic profiles of target customer groups, together with the most appropriate approach to these groups.
A variety of different analytics can be applied as appropriate.
Typical applications include detection of customer records containing potentially falsified data, identifying behaviour patterns that typically precede attempted fraud, identifying people and places that a fraudulent person is associated with, and raising alerts when known or potentially fraudulent people contact your organization.
The characteristics of your data will determine the most appropriate.
There is a wide range of available analytics, from simple checking of data values to identify invalid, contradictory, or missing values, through to the identification of deliberately falsified data, the identification of missing or duplicate records, and the matching of records from different data sources.
SuperSTAR offers true ad hoc analytics as data is not summarised prior to tabulation and only minimal transformations are made to the data during loading. This means you have access to all of the data in its original state.
Behind the scenes the advanced SuperSTAR tabulation engine, SuperSERVER, employees OLMAP technology to ensure query response times are fast, even on millions of rows of micro data.
Navigation and querying the data is easy with SuperSTAR's intuitive interfaces - SuperCROSS and SuperWEB.
The interfaces are designed to make it easy to ask the types of ad hoc questions you most often want to ask.
Please contact us to discuss your requirements.