a) Data Plots

Analysis

Trend

There is a general upwards trend of turnover in the data, which the autoplot clearly shows from 2014 to late 2022, however after this point, the upwards trend appears to be stopping,however overall the general upwards trend from 2014 shows that growth has been steadily occuring. This can also been seen from the lag plot having a fairly linear pattern indicating a steady trend.

Seasonality

We can see a very clear example of seasonality from the autoplot and the seasonal plot, with turnover typically having its up and down cycles within the year.

Cyclicity

This dataset does not appear to indicate any cyclic behaviour, as throughout the whole time series there has been nearly only upward growth.

Seasonal Patterns

Seasonally, online non-food retailing usually reaches its peak around November, which can be seen in all of the plots, meaning an increase in consumer spending during this time, which is likely due to the large amount of sales surrounding Black Friday in November. This can also be seen with the lag plots, with November generally being an outlier. A release by the ABS (available at https://www.abs.gov.au/media-centre/media-releases/black-friday-sales-lift-retail-turnover-november), in 2023 indicates large increases in monthly turnover in November in the non-food retail sector overall, up to an 8.5% pt increase from Oct-23 to Nov-23 for household goods, which is likely to extend to the online non-food retail sector.

Unusual Years

2023 represents an unusual year, as it is the first year to not show a large growth in turnover. In terms of what this indicates, it could indicate a decline in consumer interest in online non-food retailing, which could be caused by a number of factors, ranging from inflation decreasing overall spending, reduced marketing on retailer’s ends, or a loss of interest in online retailing specifically. The Australian Ecommerce Report 2023 (avalaible at https://m.iabaustralia.com.au/asset/466:iab-pureprofile-australian-ecommerce-report-2023.pdf) cites an 8% increase for consumers saying “cost of living pressures mean I don’t have as much to spend on less essential retail shopping”, which could be a reason for the lowered growth in 2023. Another unusual year is 2022, with a sharp decline in consumer growth specifically around August to October, while experiencing more significant growth around May, June, and November. This could indicate cost of living pressures causing consumers to keep most non-essential retail spending confined to times of the year with bigger sales such as Black Friday.

b) Data Plots

Discussion

While these transformations have moderated the variance of the original time series, neither completely stabilized it. I think a transformation is needed as the variance of the data increases alot with the level of turnover, thus a transformation is helpful in order the stabilize the variance, as this would allow for easier and more effective analysis.