1/115
Looks like no tags are added yet.
business analytics
exploring and analyzing a large volume of data to predict the relationship or pattern of phenomena of interest and generate insights for business organizations
what are all the types of analytics
static reports, descriptive analytics, predictive analytics, prescriptive analytics, and big data analytics
static reports / interaction dashboards
database queries shown as reports and dashboards to users
descriptive analytics
describes the past status of the domain of interest using a variety of tools through techniques such as report, data visualization, dashboards, and scorecards
predictive analytics
applies statisical and computational methods and models to data regarding past and current events to predict what might happen in the future
prescriptive analytics
uses results of predictive analytics along with optimization and simulation tools to recommend actions that will lead to a desired outcome
big data analytics
analytics mainly on unstructured (non-tabular) data on a large scale
what is the first and most important step of analytics
formulating the problem
most prediction problemscan be conceptualized as
a model to transform input data to predict outcomes
supervised learning
trains a model to predict outputs from future inputs with a dataset of example inputs and outputs
what are the two methods of supervised learning
classification and regression
classification
the output variable is a category
regression
the output variable is a continuous value
what is supervised learning generally referred to as
predictive analytics
unsupervised learning
trains a model to find existing patterns in the data to learn more about ti with a dataset where only the inputs are known
what are the two methods of unsupervised learning
association and clustering
association
you want to discover rules that describe your data
clustering
you want to discover the inherent categories in the data
what is unsupervised learning generally referred to as
descriptive analytics
every model has
pros and cons
the primary purpose of prediction is to
predict new observations
a good model should capture
the underlying logic of the dataset enough to be applied to new data
underfitting
a model does not capture any underlying logic and any relationship to be found
overfitting
a model has focused on the particular data too much to be applied to new data
to avoid underfitting,
collect sufficient data and employ a sophisicated model
data
compilations of facts, figures, or other content
what are the two types of data
numerical and non-numerical
information
data that have been organizaed, analyzed, and processed in a meaningful and purposeful way
knowledge
derived from using a blend of data, contextual information, experience, and intuiton
business analytics helps
transform data, to information and knowledge, to decisions, and finally to action
population
consists of all items of interest in an analytics application
sample
a subset of the population that is representative
you use a sample to
compute a sample statistic to estimate the unknown population parameter and make inferences about the unknown population parameter
structured data
resides in a predefined row-column format that includes numerical values, dates, categories, and combinations of words and numbers
unstructured data
does not conform to a predefined row-column format with multimedia of photographs, videos, and audio
big data
massive volume of both structured and unstructured data that are extremely difficult to manage, process, and analyze using traditional data-processing tools
what are the three characteristics (or V’s) of big data
volume, velocity, variety
volume
immense amount of data compiled from a single source or multiple sources
velocity
generated at a rapid speed, managing data streams can be critical issue
variety
all types, forms, granularity and structures
discrete variable
assumes a countable number of values
continuous variable
assumed an uncountable number of values within an interval
what are the four major scales of measurement
nominal, ordinal, interval, and ratio
what are the two categorical scales of measurement
nominal and ordinal
what are the two numerical scales of measurement
interval and ratio
nominal
observations differ by name or label; all we can do is categorize or group them
ordinal
observations can be categorized or ranked
interval
rank and find meaningful differences; zero value is arbitrary and does not reflect the absence of a characteristic; ratios are not meaningful
ratio
like interval but zero is not arbitrary and ratios are meaningful
garbage data ends up yielding
garbage results despite a perfect model
raw data is almost always
inaccurate, incomplete, or erroneous, which needs to be cleaned before analysis
data cleaning
the process of identifying incomplete, incorrect, inaccurate, or irrelevant parts of the data and then replacing, modifying, or deleting the dirty or coarse data
tidy data
a way to organize tabular data with a consistent data structure
data wrangling
the process of transforming and mapping data from one “raw” data form into another format to make it more appropriate and valuable for analytical purposes
reshaping data
reorganize the values of a table into a new layout
unpivot
moves column names into a variable column, gathering the column values int oa single value column
pivot
moves the unique values of a key column into the column names, spreading the values of a value column across the new columns
textsplit or textjoin
splits or combines cells into individual, isolated values
each observation
corresponds each row
each variable
corresponds to each column
each value of a variable for an observational unit
should be stored as an element of a table
inappropriate and/or missing values could cause
unexpected results or errors
what are the two ways to statistically describe a distribution
central tendency and disperion & shape
central tendency
mean, median, mode
dispersion & shape
range, IQR, standard deviation, variance, skewness, kurtosis
the arithmetic mean
the primary measure of central location
median
the middle value of a data set
mode
the observation that occurs most frequently
measures of dispersion
gauge the underlying variability of the variable
measures of shape
reveal whether the dsitrbution of the varaible is symmetric or if the tails are more or less extreme than the normal distribution
measures of association
show whether two numerical variables have a linear relationship
range
difference between the maximum and minimum
interquartile range (IQR)
the difference between the third quartile and the first quartile
mean absolute difference (MAD)
the average of the absolute differences between the observations and the mean
variance and standard deviation
two most widely used measures of dispersion
coefficient of variation (CV)
relative measure of dispersion and adjusts for differences in the magnitudes of the means
skewness coefficient
measures the degree to which a distribution is not symmetric about its mean
kurtosis coefficient
a summary measure that tells us whether the tails of the distribution are more or less extreme than the normal distribution
covariance
measures the direction of the linear relationship
correlation coefficient
describes both the direction and strength of the linear relationship between x and y
outliers
extremely large or small observations for a variable
z-score
measures the distance of an observation from the mean in terms of standard deviations
different visualization tools help us to
visualize categorical or numerical data, single variable or relationship between multiple variables
line plot
track changes over a period of time
bar graph
comparing between different groups or track changes over time
pie chart
comparing parts of a whole
scatter plot
show the relationship between items based on two different variables
histogram
illustrates the distribution of numerical data across categories
box plot
displays a summary of the range and statistical distribution of a dataset based on a five-number summary
what are the single variable plots
histogram, box plot, and bar chart
what are the multiple variable plots
scatter plot, line graph, heatmap, bar chart
what are the numeric plots
histogram, box plot, scatter plot, line graph
what are the categorical plots
bar chart, heatmap, box plot
statistic
a variable whose value depends on the sample
estimator
a statistic used to estimate a parameter
estimate
a particular value of an estimator
confidence interval
provides a range of values that, with a certin level of confidence, contains the population parameter of interest
hypothesis testing
used to resolve conflicts between two competing hypotheses on particular population of interest
if sample evidence is inconsistent with the null hypothesis
we reject the null hypothesis
if sample evidence is not inconsistent with the null hypothesis,
then we do not reject the null hypothesis