•



•



•



•



•



•



Time data
•Continuity  - temperature, party support
•Discontuinity – elections
•
•Data: Annualy, monthly, daily, (hours, minutes, seconds)
•Iregular: elections, exams, conflicts
•
•Easy to find spurios correlation
•
•

•



Three elements of Time Data
•Trend – the overal direction of evolution
•E.g. Global warming, increasing prices
•Seasonality – regular changes in data
•Wheather, unemployment, activity during day
•White noise

Trend
•Usually the most important things
•Allows us to say what is happening
•Forecast (be cautius with that)
•The main source of  spurious correlation

•
?


Seasonality
•Usually the most anoying aspect of time data
•The solution is to look on the whole season
•The detail is lost in such case
•

•



White noise
•Short term irregular deviations
•Important when we want to see impact of some event
•
•Make the general information hard to see
•Moving average – replace current value by average of neighbouring values
•Usualy 3, 5 or 7
•Depends on data (e.g. Monthly temperature)
•

Plot of monthly temperature – line chart
•


Change of labels and content



•



•



Just for the assignment
•Error bars
•When the data are based on representative sample
•There is uncertainity around obtained number
•Standard error: interval where the true value is probably located