What is Dataconda?
Dataconda is a software that solves classification and regression problems in a relational database, as opposed to a single flat table. The user selects a class attribute contained in a table of a relational database, and Dataconda builds and selects thousands of predictors by exploring the whole database and aggregating information, without any user intervention.
For example, Dataconda may find that the best predictor for “customer value” is the amount of money spent by the customer in cheap electronics, even if the user has not built any such attribute!
Hard to believe, right? Watch the tutorial to see with your eyes.
Why is Dataconda different from other data mining software?
Whereas traditional data mining software can only analyze a flat table, Dataconda can analyze a whole relational database.
Dataconda is a software that solves classification and regression problems in a relational database, as opposed to a single flat table. The user selects a class attribute contained in a table of a relational database, and Dataconda builds and selects thousands of predictors by exploring the whole database and aggregating information, without any user intervention.
For example, Dataconda may find that the best predictor for “customer value” is the amount of money spent by the customer in cheap electronics, even if the user has not built any such attribute!
Hard to believe, right? Watch the tutorial to see with your eyes.
Why is Dataconda different from other data mining software?
Whereas traditional data mining software can only analyze a flat table, Dataconda can analyze a whole relational database.
Example
Consider, for example, this database diagram:
Consider, for example, this database diagram:
Problem
We want to find the reasons for returning a purchased product. Practically, we want to classify Purchases depending on the value of their binary (0 or 1) attribute Returned.
Solution
Dataconda generates a "mining table" with one row per purchase and one column per descriptor of purchases. The descriptors not only include attributes that are readily available through straightforward joins (such as the client's gender and the product price), but they will also include complex aggregate information, such as:
- The current client's past number of returns
- The number of clients younger than 45 who have already bought the current product
- The average price of products returned by the current client in the past
- ...and thousands of other descriptors...
Watch the tutorial to learn more.