| The important processes that have to be clearly | | | | will be kept for different periods. In some cases, |
| delineated for Data Mining, Analysis and Modelling are: | | | | aggregated data may be kept rather than source |
| Data model: what data will be available and how will it | | | | data. All of these factors effect the data modelling |
| flow? | | | | exercise and the eventual modelling software |
| Data gathering: how will data be gathered both in | | | | requirements. |
| physical and technological terms? | | | | Turning the data into useful information requires: |
| Data gathered: what data will be gathered? | | | | Identifying the issue(s) |
| Data types: what types of data will be gathered? | | | | Assembling the data set(s) |
| Data formatting: how will data be held? | | | | Building models |
| Data warehousing: where will data be held? | | | | Verify models |
| Data mining: how will we retrieve data from the | | | | Interpretation of the results |
| warehouse? | | | | Automation of the delivery |
| Information modelling: how will we create models and | | | | Thereafter, modelling tools and techniques have to |
| what of? | | | | be used. These can be divided into two groups: |
| Information access: how will we access the data | | | | theory driven and data driven. |
| models and reports? | | | | Theory driven modelling (hypothesis testing) |
| Presentation & reporting: on what will we report? | | | | attempts to substantiate or disprove preconceived |
| Most companies want to know essential information | | | | ideas. Theory driven modelling tools require the user |
| about customers at every point of contact, for | | | | to specify most of the model based on prior |
| example: | | | | knowledge and then tests to see if the model is valid. |
| Lifetime value | | | | Data driven modelling tools automatically create the |
| X sell and upgrade potential | | | | model based on patterns they find in the data. This |
| Acquisition cost | | | | also needs to be tested before it can be accepted |
| Channel preferences | | | | as valid. |
| Loyalty/retention | | | | Modelling is an iterative process with the final model |
| Purchase behaviour patterns | | | | usually being a combination of prior knowledge and |
| Much of the data that they have will have different | | | | newly discovered information. |
| frequencies of change, refreshment or occurrence. It | | | | |