info mgmt quiz 2 bcor 2205 exam with |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
correct answers |||\\\
analysis gap - correct answer✔✔the space between data and information
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
dirty data - correct answer✔✔duplicate, incorrect, missing
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
blenders - correct answer✔✔combinations of models |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
deep learning - correct answer✔✔subset of ml and ai; based off neural networking (the
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
human brain) |||\\\
sme - correct answer✔✔subject matter expert (i.e., 10kdiabetes: doctors/nurses)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
unit of analysis - correct answer✔✔the "who" or "what" we are studying (e.g., patients for
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
diabetes)
prediction target - correct answer✔✔what we are trying to predict (unit of analysis must be at |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
this level; e.g., patients are the unit of analysis, and whether or not they were readmitted is
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
the target)
|||\\\
cross-validation - correct answer✔✔looking at different pieces of the data to minimize error |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
(use logloss in datarobot; striving for the smallest #)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
why cross-validation is important - correct answer✔✔to get a better sense of the prediction
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
ability of the algorithm because one validation isn't always representative
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
, steps of cross-validation - correct answer✔✔1. identify holdout (usually 20%)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
2. the rest of the data is used for training and validation (usually 80%)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
(e.g., if there are 100 rows in a 5-fold validation, 20 rows will be used in the holdout, and the
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
rest of the data will be split into 5 parts, so 16 rows per part)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
target leakage - correct answer✔✔- a feature you wouldn't have access to at the time (e.g.,
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
measure amount of inches of rain when trying to predict rainy days) |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
- the feature and the target are highly correlated (e.g., asking which antibiotics a patient is
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
taking when trying to figure out whether or not they have pneumonia)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
types of targets - correct answer✔✔classification|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
binary/boolean
continuous/regression
classification (types of targets) - correct answer✔✔e.g., what color is the light? green, yellow, |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
green
binary/boolean (types of targets) - correct answer✔✔e.g., yes/no |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
continuous/regression (types of targets) - correct answer✔✔e.g., how much $ is the house |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
worth?
etl - correct answer✔✔extract (get data)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
transform (convert into nice format) |||\\\ |||\\\ |||\\\ |||\\\
load (put clean data in final location)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
internal data - correct answer✔✔data from within your organization |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
correct answers |||\\\
analysis gap - correct answer✔✔the space between data and information
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
dirty data - correct answer✔✔duplicate, incorrect, missing
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
blenders - correct answer✔✔combinations of models |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
deep learning - correct answer✔✔subset of ml and ai; based off neural networking (the
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
human brain) |||\\\
sme - correct answer✔✔subject matter expert (i.e., 10kdiabetes: doctors/nurses)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
unit of analysis - correct answer✔✔the "who" or "what" we are studying (e.g., patients for
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
diabetes)
prediction target - correct answer✔✔what we are trying to predict (unit of analysis must be at |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
this level; e.g., patients are the unit of analysis, and whether or not they were readmitted is
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
the target)
|||\\\
cross-validation - correct answer✔✔looking at different pieces of the data to minimize error |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
(use logloss in datarobot; striving for the smallest #)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
why cross-validation is important - correct answer✔✔to get a better sense of the prediction
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
ability of the algorithm because one validation isn't always representative
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
, steps of cross-validation - correct answer✔✔1. identify holdout (usually 20%)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
2. the rest of the data is used for training and validation (usually 80%)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
(e.g., if there are 100 rows in a 5-fold validation, 20 rows will be used in the holdout, and the
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
rest of the data will be split into 5 parts, so 16 rows per part)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
target leakage - correct answer✔✔- a feature you wouldn't have access to at the time (e.g.,
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
measure amount of inches of rain when trying to predict rainy days) |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
- the feature and the target are highly correlated (e.g., asking which antibiotics a patient is
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
taking when trying to figure out whether or not they have pneumonia)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
types of targets - correct answer✔✔classification|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
binary/boolean
continuous/regression
classification (types of targets) - correct answer✔✔e.g., what color is the light? green, yellow, |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
green
binary/boolean (types of targets) - correct answer✔✔e.g., yes/no |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
continuous/regression (types of targets) - correct answer✔✔e.g., how much $ is the house |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
worth?
etl - correct answer✔✔extract (get data)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
transform (convert into nice format) |||\\\ |||\\\ |||\\\ |||\\\
load (put clean data in final location)
|||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
internal data - correct answer✔✔data from within your organization |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\ |||\\\
