DATA WAREHOUSING & MINING EXAM
(CH #5) QUESTIONS WITH VERIFIED
ANSWERS
Knowledge Discovery in Databases (KDD) - Answer-The process of using data
mining methods to find useful information and patterns in the data; A comprehensive
process that encompasses data mining.
Classifications - Answer-A data mining method; a popular member of the machine-
learning techniques; learns patterns from past data in order to place new instances
into their respective groups or classes.
Regression - Answer-What is being predicted is a numeric value.
predictive accuracy - Answer-The model's ability to correctly predict the class label of
new or previously seen data.
Actual class labels of a test dataset are matched against the class labels predicted
by the model.
Accuracy Rate - Answer-The percentage of test dataset samples correctly classified
by the model.
Speed - Answer-The computational costs involved in generating and using the
model.
Robustness - Answer-The model's ability to make reasonably accurate predictions,
given noisy data or data with missing and erroneous values.
Scalability - Answer-The ability to construct a prediction model efficiently given a
rather large amount of data.
Interpretability - Answer-The level of understanding and insight provided by the
model.
e.g. How and what the model concludes on certain predictions.
Confusion Matrix - Answer-primary source for accuracy estimation; with the
horizontal axis representing correct decisions and the vertical axis representing the
errors.
True Positive Rate (Accuracy Metric) - Answer-The ratio of correctly classified
positives divided by the total positive count (i.e. hit rate)
TP / (TP+FN)
True Negative Rate (Accuracy Metric) - Answer-The ratio of correctly classified
negatives divided by the total negative count (i.e. false alarm rate)
, TN / (TN+FP)
Accuracy - Answer-The ratio of correctly classified instances (positives and
negatives) divided by the total number of instances.
(TP+TN) / (TP+TN+FP+FN)
Precision - Answer-The ratio of correctly classified positives divided by the sum of
correctly classified positives and incorrectly classified positives.
TP / (TP+FP)
Recall - Answer-Ratio of correctly classified positives divided by the sum of correctly
classified positives and incorrectly classified negatives.
TP / (TP+FN)
Simple Split - Answer-Partitions data into two mutually exclusive subsets called a
training set and a test set (or holdout set).
k-fold cross-validation - Answer-The complete dataset is randomly split into k
mutually exclusive subsets if approximately equal size.
Leave-One-Out - Answer-Method similar to the k-fold method; k takes the value of 1;
that is, every data point is used for testing once on as many models developed as
there are number of data points. (Best for small data points)
Bootstrapping - Answer-A fixed number of instances from the original data is
sampled (With replacement) for training and the rest of the dataset is used for
testing. The process is repeated as many times as desired.
Jackfiring - Answer-Similar to the leave-one-out methodology; the accuracy is
calculated by leaving one sample out at each iteration of the estimation process.
Area Under the ROC Curve - Answer-A graphical assessment technique where the
true positive rate is plotted on the Y-axis and false positive rate is plotted on the X-
axis. Determines the accuracy measure of a classifier: Value of 1 indicates a perfect
classifier whereas 0.5 indicates no better than random chance.
Case-based reasoning - Answer-Uses historical cases to recognize commonalities in
order to assign a new case in the most probable category.
Bayesian Classifiers - Answer-Uses probability theory to build classification models
based on the past occurrences that are capable of placing a new instance into a
most probable class (or category).
generic algorithm - Answer-Use of the analogy of natural evolution to build directed
search-based mechanisms to classify data samples.
(CH #5) QUESTIONS WITH VERIFIED
ANSWERS
Knowledge Discovery in Databases (KDD) - Answer-The process of using data
mining methods to find useful information and patterns in the data; A comprehensive
process that encompasses data mining.
Classifications - Answer-A data mining method; a popular member of the machine-
learning techniques; learns patterns from past data in order to place new instances
into their respective groups or classes.
Regression - Answer-What is being predicted is a numeric value.
predictive accuracy - Answer-The model's ability to correctly predict the class label of
new or previously seen data.
Actual class labels of a test dataset are matched against the class labels predicted
by the model.
Accuracy Rate - Answer-The percentage of test dataset samples correctly classified
by the model.
Speed - Answer-The computational costs involved in generating and using the
model.
Robustness - Answer-The model's ability to make reasonably accurate predictions,
given noisy data or data with missing and erroneous values.
Scalability - Answer-The ability to construct a prediction model efficiently given a
rather large amount of data.
Interpretability - Answer-The level of understanding and insight provided by the
model.
e.g. How and what the model concludes on certain predictions.
Confusion Matrix - Answer-primary source for accuracy estimation; with the
horizontal axis representing correct decisions and the vertical axis representing the
errors.
True Positive Rate (Accuracy Metric) - Answer-The ratio of correctly classified
positives divided by the total positive count (i.e. hit rate)
TP / (TP+FN)
True Negative Rate (Accuracy Metric) - Answer-The ratio of correctly classified
negatives divided by the total negative count (i.e. false alarm rate)
, TN / (TN+FP)
Accuracy - Answer-The ratio of correctly classified instances (positives and
negatives) divided by the total number of instances.
(TP+TN) / (TP+TN+FP+FN)
Precision - Answer-The ratio of correctly classified positives divided by the sum of
correctly classified positives and incorrectly classified positives.
TP / (TP+FP)
Recall - Answer-Ratio of correctly classified positives divided by the sum of correctly
classified positives and incorrectly classified negatives.
TP / (TP+FN)
Simple Split - Answer-Partitions data into two mutually exclusive subsets called a
training set and a test set (or holdout set).
k-fold cross-validation - Answer-The complete dataset is randomly split into k
mutually exclusive subsets if approximately equal size.
Leave-One-Out - Answer-Method similar to the k-fold method; k takes the value of 1;
that is, every data point is used for testing once on as many models developed as
there are number of data points. (Best for small data points)
Bootstrapping - Answer-A fixed number of instances from the original data is
sampled (With replacement) for training and the rest of the dataset is used for
testing. The process is repeated as many times as desired.
Jackfiring - Answer-Similar to the leave-one-out methodology; the accuracy is
calculated by leaving one sample out at each iteration of the estimation process.
Area Under the ROC Curve - Answer-A graphical assessment technique where the
true positive rate is plotted on the Y-axis and false positive rate is plotted on the X-
axis. Determines the accuracy measure of a classifier: Value of 1 indicates a perfect
classifier whereas 0.5 indicates no better than random chance.
Case-based reasoning - Answer-Uses historical cases to recognize commonalities in
order to assign a new case in the most probable category.
Bayesian Classifiers - Answer-Uses probability theory to build classification models
based on the past occurrences that are capable of placing a new instance into a
most probable class (or category).
generic algorithm - Answer-Use of the analogy of natural evolution to build directed
search-based mechanisms to classify data samples.
