for Business Intelligence-I need you write Part b Task B (section a-c)

Tourism is a key component of Australia’s economy. From historical data, forecasts can be made on the number of tourists in Australia in order to estimate and plan accordingly. Public data on outbound tourists are available to estimate the number of tourists from abroad. This report uses 10-year monthly data of short-term arrivals to predict the number of tourists in 12 months. With different time series techniques under consideration, this report chooses Winters’s method to predict the number of tourists from July-2017 to June-2018.

This report attempts to find a proxy time series that can be the best to be used to estimate the number of tourists in Australia in the coming 12 months. This report will introduce the rationale behind the choice of the time series. Patterns found in time series will be discussed in next section followed by introduction of different smoothing approaches to the time series. The last part of the report will outline the chosen time series forecasting method used to predict the time series in the coming 12 months.

This session discusses the rationale behind the choice of the series used to predict the number of tourists visiting Australia from July-2017 to June-2018. The secondary data available is from Australian Bureau of Statistics. The time series used in this analysis is the short-term visitor arrivals in Australia. The time frame of the historical data collected is the monthly data from Jan-2007 to Dec-2016. A total of 120 data points are collected.

A graph of Australia previous new visitors is plotted against the respective time in months (years). The aim of this graphical display is to generate a visible trend which would in return produce an insight of the expected possible trend of Australia tourism industry. The data used was for all short-term visitors, as long-term visitors would be a biased data to project the actual trend in the tourism industry. To understand the trends clearly, the time series data are sub-divided into four broad components namely; Irregular, cyclic, trend and finally seasonal time series. All of these components are very important in projecting, understanding and defining the nature of the trend observed.

Short-term visitors are mainly people on business trips, school trips, and also person on vocational travel due to weather changes or events such as weddings. A plot of this data generates a seasonal time series graph with fluctuations of either cyclic or trend time series. The graph should adapt a directly or inversely proportional line curve depending on the nature of Australia tourism industry outputs.

The time series plot below displays the short-term visitor arrivals from Jan-07 to Dec-16

In the graph output above, it can be observed that seasonal time series line curve is generated. In the initial stages, the graph is observed to be constant with minimal cyclic movements’ to a certain point where the line curve raises steadily- which is brought about by a constant increase in visitors population at different periods of time.

It can be observed initially that a population of 450000 visits in Australia is reflected in the month February in the year 2006, while in the end of the seasonal trend, the graph simply indicating a population of 720000 visitors in Australia in the period of January, 2017. The seasonal curve is seen to be oscillating (moving up-and-down) steadily in the beginning, indicating a time series component referred to as “Cyclic Time Series”.

The graph is observe to have constant movements between the period January-February, 2006 and April, 2012 after which a rapid increase is observed between April, 2012 all the way to the end. The constant movement might be due to factors such as bad politics, unavailable visa and also unstable economy.The oscillation movements may be due to un-even tourist visits in the country. The numbers keep on increasing and decreasing with time, no inconsistency of growth. The rapid increase would be due to enhanced tourism policies, quick approvals of visa, suitable immigration process and also stable political aspects in Australia making it a conducive tourist attraction area.

• Tourists spots of Australia

These sites have drawn people from different corners of the world, who visits Australia to be part of the community and also to enjoy the beautiful sceneries. This has boosted the country tourism by double digits in the last two years as depicted by the seasonal time series graph above.

• Economy aspect

Australia has one of the best developed economies in the whole world. The developed economy is inducing Australia as a business site for learning new trends and also a world catchment area for investors. Tourism is one of the main industries that raise the economic status of a country. Tourism does not only bring revenue to a country but it also creates job opportunities to residences.

• Immigration

Movement of people from one country to another is known as Immigration. Australia favorable immigration process and documents has highly helped in increasing visits in the land. More people from different side of the world are finding refuge in the country’s favorable policies, beautiful sceneries and peaceful atmosphere.

• Political aspect of Australia

Political Stability is another aspect that can impose a huge impact in economy of a country and also the movement of people in-and-out of a country. Australia political stability has facilitated numerous growth of the tourism industry over the past two years. Healthy politics does not only harmonize the citizens of a country, but also the relationship with the entire world.

• Visa accessibility

Quick and easily accessibility of visa is another favorable aspect of movement that boosts travelling. Australia has accessible visa from the different countries, this has enhanced new visitors in the country from different countries. Visa accessibility and approval policy for this country has in return boosted the rate of new visits in Australia.

Weather may have placed an impact on the seasonal trend pattern observed. Many tourists are attracted by natural scenery in Australia. However, these tourist spots may be better visits in certain seasons of the year, leading a seasonal trend of foreign tourists. On the other hand, many tourists can come to Australia with limited time frame during the year. It is expected that such seasonal pattern will continue in future periods.

A friendlier visa policy to foreign tourists by Australian government will lead to a better growth of tourists, which affects the slope of the trend of the time series. It is difficult to make predictions on the continuation of friendly visa policy. For the time frame of our prediction, it should be expected it is less likely to change under the current government. Thus, it is also expected that the trend will continue in the forecast period.

The time series can be smoothened by three different methods. The first method used is simple exponential smoothing. The smoothing parameter is optimized using Minitab. The smoothing parameter is 0.158. The smoothing plot for the time series is shown below. However, it is found the single exponential method cannot cope with the seasonal pattern well, as indicated by fits indicated below. This smoothing method is suitable for time series without trend or seasonal component.

Another method used is double exponential smoothing method. It can cope with trend and seasonal pattern of time series. The trend component and level component are separately smoothened. The parameter used for trend and level component based on build-in optimization package in Minitab are 0.018 and 0.872 respectively. From the smoothing plot below, it can be found that the fits can map the seasonal component of the time series.

The third smoothing method used in this report is Winters multiplicative model. The model equation can be found in appendix. The parameters used in fitting the Winters model are 0.2, 0.2 and 0.2 for seasonal, trend and level components. The smoothing plot for the time series is shown as below

From the above analysis, the choice of the forecasting method used is Winters multiplicative model. The decision is made based the in-sample forecast MAD and MAPE. Both metric show that the Holt-Winters method give a better fit to the time series within the sample. On the other hand, from the fitting plot, it can be found that Holt-Winters model fit the time series better. The seasonal length in the Holt-Winters model has to be set at 24 in order to provide forecasts with more than 12-months horizon. Historical data is only available up to December 2016. In order to make forecasts up to June 2018, it is required to adjust the seasonal length setting of the model to 24 so as to make forecasts for up to 24 months.

The forecasts are made as below

This report analyzes the time series approach used to predict the number of tourists in Australia in the next 12 months from Jul-2017 to Jun-2018. The short-term visitor arrival is the closest proxy available from ABS to predict the number of tourists. Three smoothing methods are used to smoothen the series. It is found the Holt-Winter multiplicative model fits the best. It is expected that seasonal pattern of the time series will persist in the future period as factors affecting the number of tourists, such as weather and holidays, is not going to be changed in near term.

Imdadullah, M. (2014, January 04). Time Series Analysis and Forecasting. Retrieved April 11, 2017, from http://itfeature.com/time-series-analysis-and-forecasting/time-series-analysis-forecasting

Mihai, D 2012, 'The Tourism Market of Australia – A Model of Managerial Performance in Running an Exotic Tourist Destination', Journal Of Knowledge Management, Economics And Information Technology, Vol 2, Iss 6, Pp 178-201 (2012), 6, p. 178

Australian Bureau of Statistics, (2017). 3401.0 - Overseas Arrivals and Departures, Australia, Feb 2017. [online] Abs.gov.au. Available at: http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3401.0Feb%202017?OpenDocument [Accessed 15 Apr. 2017].

Fitting using double smoothing model

Fitting equation using simple exponential smoothing

where is the smoothing parameter

Making forecasts using Holt-Winters multiplicative model

10