Term
what can go wrong in data mining?
which one is worse? |
|
Definition
learning things that aren't true
patterns may not represent any underlying value.
learning things that are true but not useful.
Data may be at wrong level of detail
#1 is worse. |
|
|
Term
| what are the data mining styles. describe them. |
|
Definition
Hypothesis testing - answer questions or gain understanding
directed - construct a model that explains or predicts one or more target variables
undirected - find overall patterns that are not tied to a particular target |
|
|
Term
| Why is hypothesis testing hard with long held beliefs? |
|
Definition
| data reflects whatever assumptions have been made in the past |
|
|
Term
| What are the hypothesis tests? |
|
Definition
test/control
A/B tests
champion/challenger tests |
|
|
Term
| for hypothesis testing, what is a test and control |
|
Definition
test group and control group (treatments)
choose overall group and randomly divide it into test/control groups
any significant difference between them can be attributed to the treatment |
|
|
Term
| for hypothesis testing, what is an A/B test |
|
Definition
compares 2 or more treatments
determines effect of minor changes
associated with direct marketing and web based retailing
paired tests |
|
|
Term
| for hypothesis testing, what is a champion/challenger test |
|
Definition
common form of A/B testing: compares new treatment (challenger) to existing treatment (champion)
new model isn't adopted until it is proven better than the old one |
|
|
Term
| Directed data mining focuses on ___ variables that are targets |
|
Definition
|
|
Term
| ___ looks for patterns that explain the target values |
|
Definition
|
|
Term
| When is predictive modeling used |
|
Definition
| when the target comes from a timeframe later than the inputs |
|
|
Term
| when is profile modeling used |
|
Definition
| when the target and inputs come from the same timeframe |
|
|
Term
| what is the goal of undirected data mining |
|
Definition
| to find overall patterns, then identify if those patterns are useful |
|
|
Term
| What are the data mining tasks |
|
Definition
Preparing data for mining
exploratory data analysis
binary response modeling
classification of discrete values & predictions
Estimation of numeric values
finding clusters and associations
applying a model to new data |
|
|
Term
|
Definition
|
|
