Data Science Final Exam Questions with 100%
Verified Answers
The two major types of data reduction are - ✔️✔️Dimensionality reduction and
numerosity reduction (the number of variables and the number of points)
Which of the following are methods of dimension reduction? - ✔️✔️-Feature
selection
-Feature extraction
-Forward selection and backward selection
-Attribute relevance analysis (e.g. information gain)
Which of these describes the process of building an using a classifier? - ✔️✔️Data
examples are fed to a classification algorithm which produces a model that can be
applied to classify other data examples.
Which of the following are considerations in classification? - ✔️✔️-Sensitivity to
noisy data or outliers
-How long it takes to build the model and how long it takes to classify an instance
-Readability of the result to a human
-Scalability of the model with big data
-Accuracy of the model on new data
-Difficulty of including both the data points and the data dimensions in the
analysis
, How do you classify an instance using a decision tree? - ✔️✔️Start at the root. Each
node represents one data attribute (aka variable, feature). Each branch
represents a value of that attribute. Follow the branch corresponding to the value
of the data point in the attribute of the node. The result node is either a label or
another attribute where you repeat the process.
Which of the following are true about the process of building a decision tree? -
✔️✔️-The process is recursive: it is built one node at a time and the algorithm is
applied at each node the same way.
-We talk about which data points are 'at a node' during this process because as
we build down the tree, for each new node we only look at data that have the
attribute values specified by the nodes that have come above (in the path from
root to current node)
-At each node, the algorithm decides which attribute to use for a new split based
on some criteria evaluated for each possible attribute.
Which of these described the GINI criteria for deciding on what attribute a new
node should use? - ✔️✔️-To pick the attribute to split for a node, we evaluate each
attribute with the GINI score. We calculate GINI(attribute) as a weighted average
of the GINI(node) score for each node that would be created by choosing that
attribute.
-Lower GINI scores are better.
Node 1: C1: 0 C2: 6
Node 2: C1: 2 C2: 4
-Match the score to the metric. Note, this is the score on the individual node, not
the score on the attribute.
A:Node 1: 0
Node 2: 0.92
Verified Answers
The two major types of data reduction are - ✔️✔️Dimensionality reduction and
numerosity reduction (the number of variables and the number of points)
Which of the following are methods of dimension reduction? - ✔️✔️-Feature
selection
-Feature extraction
-Forward selection and backward selection
-Attribute relevance analysis (e.g. information gain)
Which of these describes the process of building an using a classifier? - ✔️✔️Data
examples are fed to a classification algorithm which produces a model that can be
applied to classify other data examples.
Which of the following are considerations in classification? - ✔️✔️-Sensitivity to
noisy data or outliers
-How long it takes to build the model and how long it takes to classify an instance
-Readability of the result to a human
-Scalability of the model with big data
-Accuracy of the model on new data
-Difficulty of including both the data points and the data dimensions in the
analysis
, How do you classify an instance using a decision tree? - ✔️✔️Start at the root. Each
node represents one data attribute (aka variable, feature). Each branch
represents a value of that attribute. Follow the branch corresponding to the value
of the data point in the attribute of the node. The result node is either a label or
another attribute where you repeat the process.
Which of the following are true about the process of building a decision tree? -
✔️✔️-The process is recursive: it is built one node at a time and the algorithm is
applied at each node the same way.
-We talk about which data points are 'at a node' during this process because as
we build down the tree, for each new node we only look at data that have the
attribute values specified by the nodes that have come above (in the path from
root to current node)
-At each node, the algorithm decides which attribute to use for a new split based
on some criteria evaluated for each possible attribute.
Which of these described the GINI criteria for deciding on what attribute a new
node should use? - ✔️✔️-To pick the attribute to split for a node, we evaluate each
attribute with the GINI score. We calculate GINI(attribute) as a weighted average
of the GINI(node) score for each node that would be created by choosing that
attribute.
-Lower GINI scores are better.
Node 1: C1: 0 C2: 6
Node 2: C1: 2 C2: 4
-Match the score to the metric. Note, this is the score on the individual node, not
the score on the attribute.
A:Node 1: 0
Node 2: 0.92
