CAP Study Guide Flashcards
(101 cards)
1
Q
What are the seven CAP domains?
A
- Frame the business problem, 2. Frame the analytics problem, 3. Data, 4. Select methodology / approach, 5. Build model, 6. Deploy solution, 7. Model lifecycle
2
Q
The five E’s are
A
ethics, education, experience, examination, and effectiveness
3
Q
One popular way to frame a business opportunity or problem is to obtain reliable information on
A
the five W’s: who, what, where, when, and why
4
Q
Five W’s: Who
A
are the stakeholders who satisfy one or more of the following with respect to the project: funding, using, creating, or affected by the project’s outcome?
5
Q
Five W’s: What
A
problem/function is the project meant to solve/perform?
6
Q
Five W’s: When
A
When: does the problem occur, or function need to be performed? When does the project need to be completed?
7
Q
Five W’s: Where
A
does the problem occur? Or where does the function need to be performed? Are the physical and spatial characteristics articulated?
8
Q
Five W’s: Why
A
does the problem occur, or function need to occur?
9
Q
After the initial analysis, it may be necessary to
A
refine the problem statement to make it more accurate, more appropriate to the stakeholders, or more amenable to available analytic tools/methods.
10
Q
In framing the analytics problem, one danger we’re trying to avoid is
A
“anchoring.”
11
Q
What is “anchoring”?
A
People have a tendency to hang on to views that they’ve seen and held before, even if they are incorrect.
12
Q
How can you help mitigate the anchoring effect?
A
Remind team that assumptions are initial and preliminary, rather than finalized views.
13
Q
Decomposition
A
the act of breaking down a higher-level requirement to multiple lower-level requirements
14
Q
A requirement should be
A
unitary (no conjunctions such as and, but, or or), positive, and testable
15
Q
What is EDA?
A
Exploratory data analaysis
16
Q
DBSCAN stands for
A
Density-based spatial clustering of applications with noise
17
Q
DBSCAN is a _____-based _____
A
density-based clustering algorithm
18
Q
DBSCAN is one of the _____ algorithms
A
most common clustering algorithms
19
Q
DBSCAN works by
A
grouping together points that are closely packed together (points with many nearby neighbors) and marks as outliers points that lie alone in low-density regions (whose nearest neighbors are too far away).
20
Q
R squared is a statistic that will give some information about
A
the goodness of fit of a model.
21
Q
In regression, the R squared coefficient of determination is a statistical measure of
A
how well the regression line approximates the real data points.
22
Q
R squared is also known as
A
Coefficient of determination
23
Q
An R squared of 1 indicates that
A
the regression line perfectly fits the data.
24
Q
Low R-squared values are
A
not always bad and high R-squared values are not always good!
25
In a normal distribution, _____ percent of the data values are within one standard deviation of the mean.
68%
26
In a normal distribution, _____ percent of the data values are within two standard deviations of the mean.
95%
27
In a normal distribution, _____ percent of the data values are within three standard deviations of the mean.
99.70%
28
Conjoint analysis
is a statistical technique used in market research to determine how people value different attributes (feature, function, benefits) that make up an individual product or service.
29
Goodness of fit
degree of linear correlation of variables, it is computed with the statistical methods such as chi-square test or coefficient of determination
30
R-squared =
Explained variation / Total variation
31
R-squared is always between
0 and 100%
32
adjusted R-squared is a modified version of R-squared that has been adjusted for
the number of predictors in the model.
33
When you add a predictor to a model, the R-squared
increases, even if due to chance alone. It never decreases.
34
If a model has too many predictors and higher order polynomials, it begins to
model the random noise in the data. This condition is known as overfitting.
35
The predicted R-squared indicates
how well a regression model predicts responses for new observations. This statistic helps you determine when the model fits the original data but is less capable of providing valid predictions for new observations.
36
In time series analysis, the Box–Jenkins method
applies autoregressive moving average ARMA or ARIMA models to find the best fit of a time-series model to past values of a time series.
37
ARMA or ARIMA
autoregressive moving average
38
Canopy Clustering
is a very simple, fast and surprisingly accurate method for grouping objects into clusters. All objects are represented as a point in a multidimensional feature space. The algorithm uses a fast approximate distance metric and two distance thresholds T1 > T2 for processing.
39
Canopy clustering is often used as preprocessing step for
the K-means algorithm or the Hierarchical clustering algorithm.
40
Canopy clustering is intended to speed up clustering operations on
large data sets
41
Hierarchical clustering involves creating clusters that have
a predetermined ordering from top to bottom.
42
In the hierarchical clustering divisive method
we assign all of the observations to a single cluster and then partition the cluster to two least similar clusters. Finally, we proceed recursively on each cluster until there is one cluster for each observation.
43
In the hierarchical clustering agglomerative method
we assign each observation to its own cluster. Then, compute the similarity (e.g., distance) between each of the clusters and join the two most similar clusters. Finally, repeat steps 2 and 3 until there is only a single cluster left.
44
subjective probability
summary measure of an individual’s beliefs about whether an event occurs
45
stratified random sampling
Intentionally sampling from subpopulations to reduce sampling error for low frequency groups
46
When the variable is ratio scale use
Box-Cox transformations
47
Blue, red, white is an example of
Categorical
48
Fully agree, partially agree, neutral, partially disagree, fully disagree is an example of
Likert-type
49
Very hard, somewhat hard, OK, somewhat easy, very easy is an example of
Semantic diferential
50
Rate factors in order of importance is an example of
Rank-order
51
30 degrees, 40 degrees, 50 degrees is an example of
Interval
52
Data quality check for Completeness:
Are all the fields of the data complete?
53
Data quality check for Correctness:
Is the data accurate?
54
Data quality check for Consistency:
Is the data provided under a given field and for a given concept consistent with the definition of that field and concept?
55
Data quality check for Currency:
Is the data obsolete?
56
Data quality check for Collaborative:
Is the data based on one opinion or on a consensus of experts in the relative area?
57
Data quality check for Confidential:
Is the data secure from unauthorized use by individuals other than the decision maker?
58
Data quality check for Clarity:
Is the data legible and comprehensible?
59
Data quality check for Common Format:
Is the data in a format easily used in the application for which it is intended?
60
Data quality check for Convenient:
Can the data be conveniently and quickly accessed by the intended user in a time-frame that allows for it to be effectively used?
61
Data quality check for Cost-effective:
Is the cost of collecting and using the data commensurate with its value?
62
Data warehouse
Staging area, centralized data, access layers (multiple OLAP data marts)
63
Data mart
Organized along a single point of view (e.g. time, product type, geography) for efficient data retrieval
64
Data mart: slice data
filtering data by picking a specific subset of the data-cube and choosing a single value for one of its dimensions;
65
Data mart: dice data
grouping data by picking specific values for multiple Dimensions
66
Data mart: drill-down/up
allow the user to navigate from the most summarized (high-level) to the most detailed (drill-down);
67
Data mart: roll-up
summarize the data along a dimension (e.g., computing totals or using some other formula);
68
Data mart: pivot
interchange rows and columns (`rotate the cube’).
69
Wrapper methods
identify a set of features on a small sample and then testing that set on a holdout sample.
70
Stochastic
Situations or models containing a random element, hence unpredictable and without a stable pattern or order. All natural events are stochastic phenomenon.
71
Often used to understand bottleneck(s) in systems
Discrete event simulation
72
Handles cases that cannot be handled by queuing theory
Discrete event simulation
73
Often used for multistage processes modeling with variations in their arrivals and service time and utilizing shared resources to perform multiple operations
Discrete event simulation
74
Designed to identify the most efficient pathway to solution; i.e., at a bank it might identify the number of tellers needed to satisfy customers in a particular time frame such as no more than 10 minutes waiting.
Queuing model
75
Monte Carlo simulation can be used if
Queuing modeling is not needed
76
Used primarily to estimate dependent variable randomness out of a set of independent variable randomness. This is especially necessary when distributions of the input variables are not necessarily normally distributed and the relationship to estimate the dependent variable is not simple (e.g. additive)
Monte Carlo simulation
77
A simulation approach used to understanding the interactions of a complex system over time.
System dynamics (SD)
78
Study of strategic decision-making processes through competition and collaboration.
Game theory.
79
The likelihood of a particular event occurring expressed as a percentage to make decisions under chosen risk or tolerance.
Probability
80
Binomial distribution
In probability theory and statistics, the binomial distribution with parameters n and p is the discrete probability distribution of the number of successes in a sequence of n independent yes/no experiments, each of which yields success with probability p. A success/failure experiment is also called a Bernoulli experiment or Bernoulli trial; when n = 1, the binomial distribution is a Bernoulli distribution.
81
Gamma distribution
distribution that arises naturally in processes for which the waiting times between events are relevant. In particular, the arrival times in the Poisson process have gamma distributions, and the chi-square distribution in statistics is a special case of the gamma distribution. Also, the gamma distribution is widely used to model physical quantities that take positive values.
82
Poisson distribution
a discrete frequency distribution that gives the probability of a number of independent events occurring in a fixed time.
83
ROC stands for
receiver operating characteristic
84
ROC curve is a
graphical plot that illustrates the performance of a binary classifier system as its discrimination threshold is varied.
85
ROC curve is created by
plotting the true positive rate against the false positive rate at various threshold settings
86
ROC true-positive rate is also known as
sensitivity in signal detection and biomedical informatics, or recall in machine learning
87
The ROC false-positive rate is also known as
fall-out
88
ROC True negative rate is also known as
Specificity
89
ROC False negative rate is also known as
Miss rate
90
ROC Type I error:
False positive
91
ROC Type II error:
False negative
92
Quadrupling the number of individuals sampled
reduces the uncertainty by half
93
Model honest assessment: For a binary target, a good practice is to ensure that you have at least
2000 observations in the smaller of the two target classes.
94
Lift
a measure of the effectiveness of a predictive model calculated as the ratio between the results obtained with and without the predictive model
95
Stochastic dominance
The concept arises in decision theory and decision analysis in situations where one gamble (a probability distribution over possible outcomes, also known as prospects) can be ranked as superior to another gamble for a broad class of decision-makers. It is based on shared preferences regarding sets of possible outcomes and their associated probabilities. Only limited knowledge of preferences is required for determining dominance.
96
Select approaches or software that deal with data accuracy at
same or better level than your data accuracy (e.g. +/- 10% if your data is less than +/- 20%)
97
If you can’t explain precisely with exact numbers you may consider using
Fuzzy logic
98
Model maintenance is necessary when
Underlying assumptions change
99
Local or neighborhood searches
take a potential solution to a problem and check its immediate neighbors (that is, solutions that are similar except for one or two minor details) in the hope of finding an improved solution. Local search methods have a tendency to become stuck in suboptimal regions or on plateaus where many solutions are equally fit.
100
Tabu search
is a metaheuristic search method employing local search methods used for mathematical optimization. Tabu search enhances the performance of local search by relaxing its basic rule. First, at each step worsening moves can be accepted if no improving move is available (like when the search is stuck at a strict local mimimum). In addition, prohibitions (henceforth the term tabu) are introduced to discourage the search from coming back to previously-visited solutions.
101
Ant colony optimization
is a probabilistic technique for solving computational problems which can be reduced to finding good paths through graphs.
