Data mining helps businesses process huge volumes of data to spot common patterns or discover new information about their customers. Dealing with this amount of big data wouldn’t be possible without automation, but marketing automation technology alone lacks the executive functioning to guide a data mining program.
Data Types
When creating tables, each column can store one type of data and that is defined during the table creation.
| Data Type | Description |
|---|---|
| Binary | Binary, length 0 to 8,000 bytes |
| Char | Character, length 0 to 8,000 bytes |
| Datetime | 8-byte datetime. Range from January 1, 1753 through December 31, 9999, with an accuracy of three-hundredths of a second |
| Image | Variable length binary data. Maximum length 2,147,483,647 |
| Integer | 4-byte integer. Value range from -2,147,483,648 through 2,147,483,647 |
| Money | 8-byte money. Range from -922,337,203,685,477.5808 through +922,337,203,685,477.5807, with accuracy to a ten-thousandth of a monetary unit. |
| Numeric | Decimal – can set precision and scale. Range -10^38 +1 through 10^38-1 |
| Smalldatetime | 4-byte datetime. Range from January 1, 1900, through June 6, 2079, with an accuracy of one minute |
| Smallint | 2-byte integer. Range from -32,768 through 32,767 |
| Smallmoney | 4-byte money. Range from 214,748.3648 through +214,748.3647, with accuracy to a ten-thousandth of a monetary unit. |
| Text | Variable-length text, maximum length 2,147,483,647 |
| Tinyint | 1-byte integer. Range from 0 through 255 |
| Varchar | Variable-length character, length 0 to 8,000 bytes |
Our data science team at Reach Marketing uses several data mining techniques to reveal insights about customers:
Cluster Detection
A kind of pattern recognition, cluster detection looks at vast data sets to see areas around which data points tend to group. These patterns are invisible at the level of individual prospect interactions, and only powerful databases are able to see them on the macro level.
Anomaly Detection
If cluster detection looks for crowds, anomaly detection looks for any data point that stands out in a crowd. By finding outliers and anomalies, our data mining experts can explore new markets or see nascent trends before the competition even knows they’re starting.
Regression
Existing data can be a powerful predictor of future outcomes. Using regression to process customer and prospect data can predict engagement, retention, sales cycle length, and more.
DATA MARTS
Your centralized data warehouse is mighty, but it must also be friendly and approachable. Data marts balance power and ease of use by narrowing the window through which users view data. When our database marketing managers develop data marts for your team, they take into account who will use them, how they will be used, and what customization options will optimize them.
What Defines a Data Mart?
Specialized
A data mart is adapted to the team that uses it, so your marketing data mart might feature lead scores and nurture programs, while your sales team’s would include direct access to the CRM.
Interconnected
Although the user-facing side of a data mart is specialized, it’s plugged directly into the main data warehouse on the back end.
Consolidated
Data structures that don’t need to be part of the larger data warehouse are stored in the data mart to filter and condense the information traveling through it.
Database Diagram
More formally, an ERD (Entity Relationship Diagram). But the relationships are not shown. They have to be explicitly defined via the SSMS interface, or programmatically. This requires PRIMARY and FOREIGN keys.
The ERD
So what do those symbols mean? They represent the number of matching records on each side
of the relationship.
Primary Keys must be unique. No value can repeat within a table. Although this is probably obvious in this case, there are many exceptions in the business world that will result in this requirement to be false.
