Time Series – from Achieved to Excellence. http:// www.nzchildren.co.nz / child_poverty.php. AS: Time Series. Using the statistical enquiry cycle to investigate time series data involves: • using existing data sets • selecting a variable to investigate
Using the statistical enquiry cycle to investigate time series data involves:
• using existing data sets
• selecting a variable to investigate
• selecting and using appropriate display(s)
• identifying features in the data and relating this to the context
• finding an appropriate model
• using the model to make a forecast
• communicating findings in a conclusion.
Investigate time series data, with justification and statistical insight involves integrating statistical and contextual knowledge throughout the statistical enquiry cycle, and may include reflecting about the process; considering other relevant variables; evaluating the adequacy of any models; or showing a deeper understanding of models.
You need to:
understand/relate to the context
research it properly and write with insight.
They need a structure to work to in order to organise your brains.
They need to be familiar with the language of statistics.
Fake it until you make it
Something you are familiar with
Note the date of the data set
“There has been increasing community, political, and education sector concern over absence from school.”
(Mallari, Loader, 2013)
Always use referencing
A national survey of state primary and secondary schools in New Zealand in 1977 (Taylor, Sturrock and White 1982) reported that the unjustified absence rate in primary schools was 0.69%, and in secondary schools it was 1.4%. Berwick-Emms (1987).
The problem of truancy is shared throughout the world (see Reid 1987, Andrews 1986). Whitney (1994:15), a British researcher, notes that ‘Truancy, like poverty, has a lengthy past history, and the two have always been closely related.
“Chronic absenteeism is most prevalent among low income students.”
The Ministry of Education survey on attendance was carried out in the week 11-15 June, 2012
The response rate was 88%
Schools recording absences on the paper form were required to make their own judgementas to whether a student was absent for all or part of a day, and whether that absence was justified based on the definitions and instructions supplied.
The survey was carried out in the week of 11-15 June 2012, close to the middle of the second school term. This week was the same week of term as the 2009 and 2011 surveys.
By analysing data from a similar time of year, factors such as winter illness would have been at broadly similar levels.
In 2012, approximately 62,000 students were absent from school for all or part of a day during the survey week. Of these, 15,000 students were unjustifiably absent from school.
What would it look like at our school?
What might be different?
Why is this Thursday lower than usual?
Referencing is necessary to avoid plagiarism.
It allows others to follow up and read what other researchers (writers) have to say about the topic.
It will become part of your
I encourage my students to use APA referencing as it is often used in university courses.
September 9, 2013
September 16, 2013
Go to the referenced sites.
that it should not be used
as a primary source for
The main problem is
the lack of authority.
Understanding and defining the problem.
Time series is essentially an investigation into ‘what has already happened and what then is likely to happen’ with consideration of how valid it all is.
Start with an overview of what they see.
Can include maximum and minimum values and average increase / decrease
Rapid/steady/gradual/plateau, increase/decrease, fallen/risen, weekly/monthly/quarterly/annual
Monthly visitor arrivals – Holiday; Jan 2000 to Oct 2012
Include numbers and gradients
“The graph shows that the trend for the number of holiday visitors was increasing from about 35000 in the beginning of 2000 up to about 59000 visitors in the beginning of 2007. This means there is a rise of approximately 300 holiday travellers every month.”
Model of good writing
“During this period, we noticed a sharp increase in the year 2000, this could be caused by multiple international events happening around that time, “Visitors to several international events - America’s Cup, APEC summit, World Netball Championship, Under-17 Soccer World Cup - contributed to this large increase” (as cited in External Migration January 2000).”
“The prominent increase in the end of 2003 could be partly contributed to the success of “The Lord of the Rings” trilogy which is completed in December 2003. This is reflected by the research “The International Visitor Survey from 2004 found that six percent of visitors to New Zealand (around 120,000 - 150,000 people) cited The Lord of the Rings as being one of the main reasons for visiting New Zealand.” (as cited in Marketing destination New Zealand through the Hobbit trilogy, 2012)”
“However, from the start of 2007 to the end of 2011, the trend remains to be relatively stable with a very slight decrease over time.”
“This change in trend could be explained by the global economical recession starting from roughly 2008, ………The change is understandable as people will first cut their budget in recreational activities like holiday travel.”
“At the beginning of 2012, especially in February, there was a sudden decrease in holiday visitors to New Zealand.”
“This could be due to a number of reasons, such as the moving holiday effect of Chinese New Year, “There were fewer arrivals from Hong Kong and China …….”
“From the estimated seasonal effect, it shows that holiday visitors are considerably higher in January and February with the peak in February at about 35,000 visitors above the trend.”
Relate back to the investigative question
“This is important for the New Zealand economy and tourism dependent industries, as that is the time where they can maximize their profits. Hence we can see that tourism industry in New Zealand is a heavily seasonally dependent market.”
“….Moreover, we notice that the peak is normally in February: this is possibly due to the fact that New Zealand is sometimes visited after going to Australia in January.”
“An increasing population of Chinese holiday visitors to New Zealand also supports the February peak, as their holiday of Chinese New Year usually starts between early and mid February. This is justified by, “Tourism is set to recover from its current slowdown due to the continuing strength of Australia and a growing Chinese market.” (as cited in Forecast commentary, 2012)”
“…. The number of visitors troughed in June (about 25000 people below the trend line) but raised slightly in July. The trough in June can be caused by the decreasing temperature as New Zealand goes to winter and the increasing amount of rainfall which makes a holidayless favourable.”
“July, however, seems to favour more visitor numbers than June; one would expect this because July is when the summer holiday of the Northern Hemisphere starts. Hence we would see an increase in holiday visitors from UK and China. This explanation is supported by…”
“…In particular, there is an outlier in the seasonality for September in 2011, which reaches to about 50,000 instead of the usual 30,000 visitors. This change is caused by positive influence brought by the Rugby World Cup of 2011.”
“After a visual inspection of the graph, the residual is relatively small with most of the variations being below 10% of the overall range (±4000) However, at the beginning of 2011, there is a large residual of about 7500. This unusual residual was probably caused by….”
Variation in data: 98000 – 21000 = 77000
Variation in Trend: 58000 – 36000 = 22000
22/77 = 0.29 i.e. 29% of the variation in the data is the trend
Variation in data: = 77000
Variation in Seasonal Effects: 35000 + 25000 = 60000
60/77 = 0.78 i.e. 78% of the variation in the data is the seasonal component
Variation in residuals = 15000
15/77 = 19%
NOTE: These are ball-park figures read off the graphs and don’t add up to 100%. The main source of variation comes from the seasonal component which contributes around 78% of the overall variation in the data.
What we are interested in is what is driving this series- in this case the seasonal component.
“After a visual inspection of the plot I am confident that the model provides a good fit as differences (white spaces) between the fitted data and the raw data are very small.”
“I predict that the average number of holiday visitors to NZ in August 2013 will be between 17400 and 44600. Hence, in the near future, my model predicts that there will be a decreasing trend in 2013.”
Take out the last 3 months of data, re-analyse and check against predictions.
“The model does not work particularly well for Sept 2012. There was an unusual decrease in visitor numbers, as opposed to the expected increase. The actual value of Sept 2012 does not even fall into the prediction interval.”
Discusses what the data does not tell you in relation to the investigative question
“…the data captures a period of economical downturn at the near end, hence predictions are generally decreasing and this will be inaccurate if the economy becomes better in the future.”
“In addition, the data only covers the total number of visitors and it does not signify the visitor spending and the length of stay. Hence it cannot give a very accurate reading of the tourism’s contribution to the New Zealand economy.”
Second analysis: Visitors of family and friends.
The student then compares the two series.
Forms a new series and discusses the contributions made and effects of key events on the new series.
You give a concise summary of the investigation which links back to the original purpose of the investigation.