Visualising time series Flashcards
what is time series analysis
Time series analysis is about visualising the quantitative relationship through
time:
2 examples of time series analysis
Business analysts hope to see profits increase over time
Government expects to see changes in relation to influential events
Time gives us a context for understanding data so that present can be
_________ and future can be _______ in light of the past
understood, predicted
what are the 6 basic time series patterns
–> Trend
–> Variability
–> Rate of change
–> Co-variation
–> Cycles
–> Exceptions
what is trend
–> A trend is the overall tendency of a series of values to increase, decrease or
remain relatively stable during a particular period of time (like 12-months)
–> Any period of time can be the basis for determining a trend but the
comparison has to be within the same period
–> a trend can be shown by using lines of best fit based on statistical models
what is variability
–> Variability is the average degree of change from one point in time to the next
throughout a span of time
what is rate of change
–> This is useful when comparing multiple series of values.
–> Percentage difference is considered
what is covariation
When 2 time series relate to one another so that changes in one are reflected as changes in the other, either immediately or later, is called co-variation.
example of covariation
–> Daily average temperature (°F) for the city
–> Daily ice cream sales (number of ice creams sold) in the city
what is cycles
Cycles are patterns that repeat at regular intervals, such as daily, weekly, monthly,
quarterly or seasonally
usually have a hockey stick pattern
what are exception
Exceptions – values fall outside the norm
what charts can be used in time series data
–> Bar charts
–> Dot plot
–> Radar charts
–> Heatmaps
–> Scatterplots (esp. in animation to analyse the correlation changes)
what is a line chart used for in time series
For analysing Patterns and Exceptions
what is a bar chart used for in time series
For Emphasizing and Comparing Individual Values
what is a dot plot used for in time series
For analysing Irregular Intervals
The dots are NOT connected in a line. Mark
each value separately.
what is a radar chart used for in time series
For comparing cycles and spot exception
radar charts are also known as
spider charts
star charts
what are heat maps used for
For analysing high volume cyclical Patterns and Exception
what are scatterplots used for in time series analysis
Animated for analysing correlation changes
what is discrete values
Discrete – values are from specific points or blocks of time
and there is a finite number of possible values
eg of discrete values
E.g. Percentage of people who pass a test each year – this value don’t change afterward
what is continuous values
Continuous – value that can be measured at any time of
day during any interval and it is constantly changing
eg of continuous values
E.g. Environment or body temperature
discrete time series data is best visualised in
Bar chart
Stacked Bar chart
Dot plot
continuous time series data is best visualised in
Line chart
Step chart
what are the 7 Techniques and Best Practices
–> Aggregating to Various Time Intervals
–> Viewing Time Periods in Context
–> Using Trend Line to enhance Perception of high-level Patterns
–> Omitting Missing Values from a display
–> Using Logarithmic Scales and Percentages to Compare Rates of Change
–> Overlapping time scales to Compare Cyclical Patterns
–> Stacking Line Charts to Compare Multiple Variables
what is aggregating
–> Time-series data can look quite
different if you change the level of
aggregation.
–> There is a need to quickly and
easily switch between various
intervals of time to examine the
data and identify the insights
what is context
–> Views of various time spans might
lead to insights that are not
available if we stick to one time
span.
–> Important to examine data in the
context of a longer period
what is the trend line
–> Take note of the time period used
for the trend line
–> Be sure to examine values that fall
outside the specified time period
and not isolated a section that
would trend quite differently if the
period is altered
–> Usually a straight line is used and
based on linear regression
–> However it may not be the best fit
so may want to consider – running average
what is the best practice for omitting
–> If missing values are found, we
can estimate the value or omit it
from displaying.
–> It is not right to treat missing value
as zero.
what is the technique for logarthmic
–> When comparing rate of changes,
important to use percentage and
logarithm scale
–> equal rate of change appear as equal slope
what is the technique of overlapping
It is easier to detect cyclical
patterns by overlapping the
multiple cycles on the same chart
what is the technique of stacking
If there is a need to compare multiple variables in the same time period but they are expressed using different units of measure, stacking line charts can be used.