I am doing a data mining project using the census bureau data set. The data set is a pdf, so I saved it as a text file, I imported as a flat source file and am trying to create my data source? But am having trouble with this? Please help?
There are no specific restrictions on the number of tables you need to use to perform data mining. You can certainly mine data in a single file/table if that contains the attributes you're interesting in mining.
For example, you could have a single customer table with demographic information like age, income and so forth and you could apply several of the available algorithms to it to derive interesting patterns - clustering to group together similar customers, decision trees to classify them based on how they behave with respect to a predictable attribute or naive-bayes to explore dependencies and differences between attributes for customers.
How would you implement the MS Time series algorithm to predict population patterns? or would you suggest I use MS decision trees?
Thanks