Introduction
History
Data Collection
Methodology
Results
Conclusion
References

Introduction

I have chosen to forecast the winner of the Eurovision song contest 2015. This is an event which receives a lot of speculation from the media and the public throughout the participating countries. I have found this a very interesting project as there are a number of variables which could affect the outcome and it is also an event which attracts the interest of lots of bookmakers. At the end of my project I have compared my results with those of the bookmakers, to see any similarities or differences.

This year marks the 60th edition of the contest and to mark the occasion, Australia has been invited to participate. This factor has also impacted my results as I do not have any previous data to include as part of my forecast. This has meant that I have been unable to assess the likely outcome of the artist they will enter. I have tried to overcome this issue by looking into the history of debuting countries and how they usually fare in their first competition, and my results of this have been included in the MODELLING section.

This topic has come with an extensive data collection task, however, there are many sources of information for public access on the internet. Once I had transferred my data into manageable excel spreadsheets, I was then able to start creating models based on the nature of the data I had collected (which too can be found in the MODELLING section of this project. I have then posted my findings in the results section, and an APPENDIX at the end is dedicated to other data I have decided to include.

History

As I have previously mentioned the Eurovision song contest has been running since 1956, making this year’s contest the 60th anniversary. The contest is an annual song competition among the member countries of the European Broadcasting Union (EBU) which was formed in 1950. This is a contradiction to the social opinion that the contest is held among solely European countries which have been invited, hence why Russia and Israel have been allowed to participate. As a non-sporting event, it is one of the most watched television shows, broadcasted in over 43 countries, and is one of the most long-running programmes in the world. Many countries that do not compete still broadcast the programme, for example United States, Japan, Argentina, New Zealand, Mexico, Brazil, South Africa, South Korea, India, Thailand and so on.

This years’ contest is to be held in Vienna, Austria, following the win of Conchita Wurst at the 2014 competition in Copenhagen, Denmark with the song “Rise like a phoenix”. The Eurovision contest offers participating artists the chance to further their careers (or launch in some cases) and many artists have found extensive fame from the competition, for example, ABBA who won the 1974 contest for Sweden with “Waterloo”. Another example is Céline Dion who won the 1988 contest for Switzerland with the song “Ne partez pas sans moi”.

The voting system has changed throughout the years, although one factor has remained constant; countries are not permitted to vote for their own entry (with the exception of the 1956 contest). Once all the countries have casted their votes, the winner is the country with the most votes. There are no prizes as such, although, trophies are usually awarded to the songwriters and the winning country is invited to become the host nation for the following contest. Up until 1998 the host country was required to provide an orchestra, however after this year CDs were allowed to be used which held backing tracks, however no voices are permitted on these tracks. In 1966 a rule was introduced which stated that countries were required to sing in one of their own official languages following an English entry in 1965 from Sweden. This restriction was continued for the next 7 years, finally ending in 1973. Subsequently, countries were allowed to perform in any language they wished. In 1977 the EBU changed the ruling back to the language restriction and this reinstated rule lasted until 1999.

The method for countries casting their votes has changed considerably throughout the years. The current system has been in place since 2009, and the voting is split 50/50; half of the countries votes come from a jury of 5 professionals from the music industry and the other 50% comes from a telephone vote (known as a jury-televote). This is a positional voting system; the country ranks all the entries and gives an allocation of points to the top 10 countries (12 points to their first choice, 10 points to the second, 8 points to the third and thus follows with the points decreasing by one unit respectively). This is a substantial change from the first contest - in 1956 a jury was made up of two members from a country and they gave 2 points to their most preferred song. In 1957 this was changed and the jury size was increased to 10 people. They had 1-10 points in which to give to their 10 favourite songs. This method continued until 1969, but changed again briefly between 1961 and 1967. Throughout the contest jury size has fluctuated, hitting a high in 1997 with twenty members. In 1998, telephone voting was introduced. Only in exceptional circumstances where there was a weak communications system was jury voting used. After 2004 semi-finals were introduced, as well as SMS voting, and then in 2009, the jury-televote was re-introduced.

In 1969 there were four winners. The EBU had no rules for a tie so all countries won. Consequently, a rule was introduced which stated that if a tie were the outcome of the results, the winning country would be the one gaining its points from the most countries. This year’s contest in Austria will see 40 countries participate however, 52 countries in total have partaken at least once.

Data Collection

Firstly, I had to decide which variables I thought would impact the outcome of the Eurovision song contest. I collected data on all of the countries ever to have participated, ranked in order of how many times they had participated, highest (with Germany at 58/59 times) to lowest (with Morocco with once). I assessed this against the number of times they had won to see if participation is a key factor in determining the outcome. After semi-finals were introduced in 2004, some countries did not qualify for the final, due to an extraodinarily large number of countries participating. In these cases, their number of participations may be higher than the number of finals they have participated in.

Country.participation.against.wins <- read.csv("~/R work files/CSV files/Country participation against wins.csv")
library("knitr", lib.loc="~/R/win-library/3.1")

## Warning: package 'knitr' was built under R version 3.1.3

kable(Country.participation.against.wins)

Country	Total.no..of.times.participated	Finals	Total.no..of.times.won
Germany	58	58	2
France	57	57	5
United Kingdom	57	57	5
Belgium	56	48	0
Switzerland	55	48	2
Netherlands	55	47	4
Spain	54	54	2
Sweden	54	53	5
Norway	53	51	3
Finland	48	44	1
Ireland	48	44	7
Austria	47	43	2
Portugal	47	40	0
Denmark	43	41	3
Italy	40	40	2
Luxembourg	37	37	0
Israel	37	31	3
Greece	35	35	1
Turkey	34	33	0
Cyprus	31	25	0
Yugoslavia*	27	27	0
Iceland	27	24	0
Malta	27	23	0
Monaco	24	21	0
Croatia	21	16	0
Estonia	20	13	1
Slovenia	20	12	0
Bosnia & Herzegovina	18	18	0
Russia	18	18	1
Poland	17	11	0
Romania	16	16	0
Lithuania	15	10	0
Latvia	15	8	1
FYR Macedonia	14	8	0
Ukraine	12	12	0
Hungary	12	10	0
Albania	11	6	0
Belarus	11	4	1
Moldova	10	8	0
Bulgaria	9	1	0
Armenia	8	7	0
Azerbaijan	7	7	1
Georgia	7	5	0
Serbia	7	5	1
Slovakia	7	3	0
Montenegro	6	1	0
Andorra	6	0	0
San Marino	5	1	0
Czech Republic	3	0	0
Serbia & Montenegro*	2	2	0
Marocco	1	1	0

I have created a pie chart, for my reader to more easily notice which coutries have won more frequently than other. To do this, I used only the countries registered to compete in May 2015, as this is the data only relevant to my forecast (although in the above table I have collected data from all the countries ever to have participated).

slices <- c(2, 1, 1, 3, 1, 1, 5, 2, 1, 7,3, 2, 1, 4, 3, 1,1, 2, 5, 2, 5)
lbls <- c("Austria", "Azerbaijan", "Belgium", "Denmark", "Estonia", "Finland", "France", "Germany", "Greece", "Ireland", "Israel", "Italy", "Latvia", "Netherlands", "Norway", "Russia", "Serbia", "Spain", "Sweden", "Switzerland", "UK")
pie(slices, labels = lbls, main="Pie Chart of Country wins")

An expectation I had when creating this table was that if practice really does make perfect, then the country which had the highest participation rate would be the country which has won most frequently. However, as the table shows, Germany has only won twice. This is contradictory to my assumption, proven by the fact that Ireland is top of the wins leaderboard with a huge 7 wins, and they have competed 10 times fewer than Germany. This then prompted me to come to the conclusion that perhaps there are other factors which affect the outcome.

There are a wide range of websites with data of previous years’ competitions readily available for public use, however none had incorporated all of the relevant variables I was looking for into one concise table, therefore I had to make a few excel spreadsheets and input all of the data myself.

every.winner.and.host.city <- read.csv("~/R work files/CSV files/every winner and host city.csv")
library("knitr", lib.loc="~/R/win-library/3.1")
kable(every.winner.and.host.city)

YEAR	HOST.CITY	WINNER	POINTS	MARGIN	RUNNER.UP	NO..OF.PARTICIPANTS	LANGUAGE.OF.WINNING.COUNTRY
1956	Lugano	Switzerland	Never announced			7	French
1957	Frankfurt	Netherlands	31	14	France	10	Dutch
1958	Hilversum	France	27	3	Switzerland	10	French
1959	Cannes	Netherlands	21	5	United Kingdom	11	Dutch
1960	London	France	32	7	United Kingdom	13	French
1961	Cannes	Luxembourg	31	6	United Kingdom	16	French
1962	Luxembourg	France	26	13	Monaco	16	French
1963	London	Denmark	42	2	Switzerland	16	Danish
1964	Copenhagen	Italy	49	32	United Kingdom	16	Italian
1965	Naples	Luxembourg	32	6	United Kingdom	18	French
1966	Luxembourg	Austria	31	15	Sweden	18	German
1967	Vienna	United Kingdom	47	25	Ireland	17	English
1968	London	Spain	29	1	United Kingdom	17	Spanish
1969	Madrid	Spain	18	No runner-up		NA	Spanish
NA		United Kingdom		-4		NA	English
NA		Netherlands				16	Dutch
NA		France				NA	French
1970	Amsterdam	Ireland	32	6	United Kingdom	12	English
1971	Dublin	Monaco	128	12	Spain	18	French
1972	Edinburgh	Luxembourg	128	14	United Kingdom	18	French
1973	Luxembourg	Luxembourg	129	4	Spain	17	French
1974	Brighton	Sweden	24	6	Italy	17	English
1975	Stockholm	Netherlands	152	14	United Kingdom	19	English
1976	The Hague	United Kingdom	164	17	France	18	English
1977	London	France	136	15	United Kingdom	18	French
1978	Paris	Israel	157	32	Belgium	20	Hebrew
1979	Jerusalem	Israel	125	9	Spain	19	Hebrew
1980	The Hague	Ireland	143	15	Germany	19	English
1981	Dublin	United Kingdom	136	4	Germany	20	English
1982	Harrogate	Germany	161	61	Israel	18	German
1983	Munich	Luxembourg	142	6	Israel	20	French
1984	Luxembourg	Sweden	145	8	Ireland	19	Swedish
1985	Gothenburg	Norway	123	18	Germany	19	Norwegian
1986	Bergen	Belgium	176	36	Switzerland	20	French
1987	Brussels	Ireland	172	31	Germany	22	English
1988	Dublin	Switzerland	137	1	United Kingdom	21	French
1989	Lausanne	Yugoslavia	137	7	United Kingdom	22	Croatian
1990	Zagreb	Italy	149	17	Ireland	22	Italian
NA					France	NA
1991	Rome	Sweden	146	0	France	22	Swedish
1992	Malmö	Ireland	155	16	United Kingdom	23	English
1993	Millstreet	Ireland	187	23	United Kingdom	25	English
1994	Dublin	Ireland	226	60	Poland	25	English
1995	Dublin	Norway	148	29	Spain	23	Norwegian
1996	Oslo	Ireland	162	48	Norway	23	English
1997	Dublin	United Kingdom	227	70	Ireland	25	English
1998	Birmingham	Israel	172	6	United Kingdom	25	Hebrew
1999	Jerusalem	Sweden	163	17	Iceland	23	English
2000	Stockholm	Denmark	195	40	Russia	24	English
2001	Copenhagen	Estonia	198	21	Denmark	23	English
2002	Tallinn	Latvia	176	12	Malta	24	English
2003	Riga	Turkey	167	2	Belgium	26	English
2004	Istanbul	Ukraine	280	17	Serbia and Montenegro	36	English
2005	Kiev	Greece	230	38	Malta	39	English
2006	Athens	Finland	292	44	Russia	37	English
2007	Helsinki	Serbia	268	33	Ukraine	42	Serbian
2008	Belgrade	Russia	272	42	Ukraine	43	English
2009	Moscow	Norway	387	169	Iceland	42	English
2010	Oslo	Germany	246	76	Turkey	39	English
2011	Düsseldorf	Azerbaijan	221	32	Italy	43	English
2012	Baku	Sweden	372	113	Russia	42	English
2013	Malmö	Denmark	281	47	Azerbaijan	39	English
2014	Copenhagen	Austria	290	52	Netherlands	37	English

This table shows data for every competition since the Eurovision was first founded in 1956. This contest was held Lugano, Switzerland and 7 countries competed in total. The winning entry from Switzerland themselves was sung in French. I have decided to include the language of the winning entry, as I have chosen to investigate whether this could affect a country’s chance of winning.

The rules concerning the languages in which countries sing their entries have changed throughout the years. From 1956 until 1965 there were no barriers to whichever language a country could choose to compete in. Following this, from 1966 to 1973 a rule was enforced which stated that countries were only permitted to sing in one of the official languages of the country participating. From 1973 to 1976, the rule was changed back and countries could once again sing in any language. From 1977 until 1998, the language restriction was imposed once more, except during the 1977 contest, Germany and Belgium were given special dispensation to sing in English due to the fact they had already gone through their own song selection phase and chose a number which they were to sing. Finally, in 1999 countries were returned their freedom to choose any lanugage to sing in. Since the ban has been lifted, countries have sung in a range of languages (usually either one of their native languages, English or a mixture of both). This is due to the fact that English is a widely spoken language throughout the world. Some countries may feel that by singing in English, they can reach a wider audience who will understand their songs.

Of all of the winning entries, I have concisely stated which languages the songs were sung in and how many times they have won:

English = 29
French = 14
Dutch = 3
Hebrew = 3
Italian = 2
German = 2
Spanish = 2
Swedish = 2
Norwegian = 2
Danish = 1
Croatian = 1
Serbian = 1

Since the language barrier was removed in 1999, English has been used by every winning country, with the exception of the 2007 contest held in Helsinki when Serbia won and chose to perform their entry in Serbian. From this analysis, it would be a valid assumption to state that singing the entry in English will definitely not damage your country’s chances of winning.

UK.points.1998.2014 <- read.csv("~/R work files/CSV files/UK points 1998-2014.csv")
France.points.1998.2014 <- read.csv("~/R work files/CSV files/France points 1998-2014.csv")
Netherlands.points.1998.2014 <- read.csv("~/R work files/CSV files/Netherlands points 1998-2014.csv")
Ireland.points.1998.2014 <- read.csv("~/R work files/CSV files/Ireland points 1998-2014.csv")
Sweden.points.1998.2014 <- read.csv("~/R work files/CSV files/Sweden points 1998-2014.csv")
plot(UK.points.1998.2014, main="Top 5 winners points 1998 to 2014", type="o", col="red",ylim=range(-5,450))
lines(France.points.1998.2014, type="o", col="blue")
lines(Netherlands.points.1998.2014, type="o", col="black")
lines(Ireland.points.1998.2014, type="o", col="green4")
lines(Sweden.points.1998.2014, type="o", col="orange")
legend("topleft",ncol=3,lty=2,col=c("red","blue","black","green4","orange"),legend=c("UK","France","Netherlands","Ireland","Sweden"))

In the above graph I have taken the data from the top winning countries, and created a line graph with a legend categorising the colours and their corresponding country. I have included Ireland, as it has won most times with 7 victories, Uk, France and Sweden follow this with 5 wins (Luxembourg is also tied with 5 wins however, has chosen not to compete since, after withdrawing from the 1994 contest for monetary reasons. Since they have only reiterated their disinterest with competing). I also decided to include the Netherlands in this graph as they are close followers with 4 victories in total.

You don’t need to be a mathematician to be able to notice that all of the values have ranged from high scores (highest recorded score on this chart is 372, which was awarded to Sweden when they won the 2012 competition) all the way throughout the range to low scores (with some very close to zero, also known as the famous “nul points”).

`Country,.EU,.native.language.and.fluency.in.english` <- read.csv("~/R work files/CSV files/Country, EU, native language and fluency in english.csv")
library("knitr", lib.loc="~/R/win-library/3.1")
Country.status.native.language.fluency <- `Country,.EU,.native.language.and.fluency.in.english`
kable(Country.status.native.language.fluency)

COUNTRY.to.compete.in.2015	NO..OF.PREVIOUS.WINS	NO..OF.TIMES.PARTICIPATED	Is.it.in.the.EU.or.EEA.	Native.language.	X..of.country.fluent.in.English
Albania	0	11	N	Albanian	0.00
Armenia	0	8	N	Armenian	10.00
Australia	0	0	N	English	97.03
Austria	2	47	Y	German	73.00
Azerbaijan	1	7	N	Azerbaijani	3.00
Belarus	0	11	N	Belarusian	0.00
Belgium	1	56	Y	Dutch	59.00
Cyprus	0	31	Y	Greek	73.00
Czech Republic	0	3	Y	Slavic	27.00
Denmark	3	43	Y	Danish	86.00
Estonia	1	20	Y	Estonian	50.00
Finland	1	48	Y	Finnish	70.00
France	5	57	Y	French	39.00
Georgia	0	7	N	Georgian	3.00
Germany	2	58	Y	German	64.00
Greece	1	35	Y	Greek	51.00
Hungary	0	12	Y	Hungarian	20.00
Iceland	0	27	EEA	Icelandic	0.30
Ireland	7	48	Y	Irish, English	98.37
Israel	3	37	N	Hebrew, Arabic	84.97
Italy	2	40	Y	Italian	34.00
Latvia	1	15	Y	Latvian	46.00
Lithuania	0	15	Y	Lithuanian	38.00
Macedonia	0	14	N	Macedonian	0.00
Malta	0	27	Y	Maltese	89.00
Moldova	0	10	N	Romanian	3.00
Montenegro	0	6	N	Montenegrin	0.00
Netherlands	4	55	Y	Dutch	90.00
Norway	3	53	EEA	Norwegian	80.00
Poland	0	17	Y	Polish	33.00
Portugal	0	47	Y	Portugese	27.00
Romania	0	16	Y	Romanian	29.00
Russia	1	18	N	Russian	5.48
San Marino	0	5	N	Italian	0.00
Serbia	1	7	N	Serbia	4.00
Slovenia	0	20	Y	Slovenian	59.00
Spain	2	54	Y	Spanish	22.00
Sweden	5	54	Y	Swedish	86.00
Switzerland	2	55	Single Market	German, French	61.28
UK	5	57	Y	English	97.74

Since having looked at previous participation rates, amount of previous wins and the language entries are sung in, I also decided to look into some other factors. The above table holds the data for all the countries to compete in the 2015 contest. I have charted the data for EU (European Union) or EEA (European Economic Area), the native language of the country and the percentage of the country that is fluent in English. Although, as I have stated in the History section, the Eurovision is not solely for countries in Europe, I was curious to see if EU membership had any impact on the results, as I considered the possibility that perhaps countries in the EU had a certain loyalty towards each other.

Ireland is my prime example, having won the most times of all the countries. Ireland is part of the EU, has English as it’s second native language and over 98% of the country are fluent in English. I doubt this is coincidental, however since the language barrier was been dropped in 1999, Ireland haven’t won. UK, Sweden, France and the Netherlands are all part of the EU, and UK, Sweden and Netherlands all have above 85% of their country fluent in English.

Methodology

In simple terms, a forecast is basically a prediction or estimate of future events or occurrences. Forecasts are often used in everyday life for example weather forecasts, how much of your wages you should save, how much food you will need from the grocery store, where to invest your money, train timetable forecasts etc. In this project I am aiming to predict the winner of the Eurovision song contest. We can also consider the effect of bias on our forecasts; we may have preconceptions about certain events or circumstances - a negative opinion of British weather may impact the level of importance we give to the weather forecasts made.

Although the aim of forecasts should be to be accurate, this is not regularly the case. The amount of variables you take into consideration when creating a forecast impacts its validity. A model based on only one variable may not be as good as one which uses three different variables. However, on the contrary, a model with 12 variables may be “over-fitting” as the variables could be too correlated. Forecasting models which are good for one dataset may not be so relevant for other datasets, so each dataset should be thoroughly examined and taken into consideration in its own right when chosing which model to fit it to.

There are some events which are useless to forecast, for example, forecasting whether a coin will land on a head or tail. Assuming we are using an unbiased coin we cannot claim that the coin will land a certain way, there is always a 50% chance of either outcome. In this project, however, I have chosen to consider a few variables which I hope remove these limitations and allow me to forecast the outcome of the ESC as accurately as possible.

There are many methods of forecasting:

The Naïve method is a method of forecasting by which all future observations are equal to the last observed value. In simple terms, the only values which are important to the forecaster are the previous observations.

Yt+h|t=Yt

The average method weights all previous observations equally. Therefore when calculating a forecast using this method, we simply sum all of the observations and divide by the total number of observations.

Yt+h|t=(1/t)??(Yt)

ETS models(Exponential Trend Smoothing models) calculate forecasts essentially by giving different weights to each observation. The reason behind this is that the most recent observations are likely to have more relevance than older observations. Therefore the most recent observations are given higher weights than the older ones. In this report I have included my Simple Exponential Smoothing outcomes in the Results section below.

ARIMA (AutoRegressive Integrated Moving Average) models “combine differencing with autoregression and a moving average model” (R. Hyndman & G. Athanasopoulos, Forecasting: Principles and Practice).

Ireland

Ireland.points.1998.2014 <- read.csv("~/R work files/CSV files/Ireland points 1998-2014.csv")
library(forecast)

## Warning: package 'forecast' was built under R version 3.1.3

## Loading required package: zoo
## 
## Attaching package: 'zoo'
## 
## The following objects are masked from 'package:base':
## 
##     as.Date, as.Date.numeric
## 
## Loading required package: timeDate

## Warning: package 'timeDate' was built under R version 3.1.3

## This is forecast 5.9

forecast.Ir <- predict(Ireland.points.1998.2014$IRELAND.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Ireland.t <- ts(Ireland.points.1998.2014$IRELAND.POINTS,start=1998,freq=1)
forecast(Ireland.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast     Lo 80    Hi 80     Lo 95    Hi 95
## 2010       36.31398 -8.341373 80.96933 -31.98049 104.6084
## 2011       36.31398 -8.341373 80.96933 -31.98049 104.6084
## 2012       36.31398 -8.341373 80.96933 -31.98049 104.6084
## 2013       36.31398 -8.341373 80.96933 -31.98049 104.6084
## 2014       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2015       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2016       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2017       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2018       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2019       36.31398 -8.341375 80.96933 -31.98049 104.6084

plot(forecast(Ireland.t),xlab="Year",ylab="Ireland Points")

UK

UK.points.1998.2014 <- read.csv("~/R work files/CSV files/UK points 1998-2014.csv")
forecast.Uk <- predict(UK.points.1998.2014$UK.POINTS,h=10)

## Warning in mean.default(x, na.rm = TRUE): argument is not numeric or
## logical: returning NA

uk.t <- ts(UK.points.1998.2014$POINTS,start=1998,freq=1)
forecast(uk.t)

##      Point Forecast     Lo 80    Hi 80     Lo 95    Hi 95
## 2015       49.32479 -17.79318 116.4428 -53.32328 151.9729
## 2016       49.32479 -17.79318 116.4428 -53.32328 151.9729
## 2017       49.32479 -17.79318 116.4428 -53.32328 151.9729
## 2018       49.32479 -17.79318 116.4428 -53.32328 151.9729
## 2019       49.32479 -17.79318 116.4428 -53.32328 151.9729
## 2020       49.32479 -17.79318 116.4428 -53.32328 151.9729
## 2021       49.32479 -17.79318 116.4428 -53.32329 151.9729
## 2022       49.32479 -17.79318 116.4428 -53.32329 151.9729
## 2023       49.32479 -17.79318 116.4428 -53.32329 151.9729
## 2024       49.32479 -17.79318 116.4428 -53.32329 151.9729

plot(forecast(uk.t))

Sweden

Sweden.points.1998.2014 <- read.csv("~/R work files/CSV files/Sweden points 1998-2014.csv")
forecast.s <- predict(Sweden.points.1998.2014$SWEDEN.POINTS,h=10)
Sweden.t <- ts(Sweden.points.1998.2014$SWEDEN.POINTS,start=1998,freq=1)
forecast(Sweden.t)

##      Point Forecast     Lo 80    Hi 80     Lo 95    Hi 95
## 2015       101.5867 22.503057 180.6704 -19.36130 222.5347
## 2016       101.5867 20.375825 182.7976 -22.61462 225.7880
## 2017       101.5867 18.272832 184.9006 -25.83087 229.0043
## 2018       101.5867 16.191855 186.9816 -29.01345 232.1869
## 2019       101.5867 14.130895 189.0425 -32.16542 235.3388
## 2020       101.5867 12.088146 191.0853 -35.28953 238.4629
## 2021       101.5867 10.061968 193.1114 -38.38830 241.5617
## 2022       101.5867  8.050867 195.1226 -41.46402 244.6374
## 2023       101.5867  6.053476 197.1199 -44.51876 247.6922
## 2024       101.5867  4.068541 199.1049 -47.55446 250.7279

plot(forecast(Sweden.t),xlab="Year",ylab="Sweden Points")

France

France.points.1998.2014 <- read.csv("~/R work files/CSV files/France points 1998-2014.csv")
forecast.Fr <- predict(France.points.1998.2014$FRANCE.POINTS,h=10)
France.t <- ts(France.points.1998.2014$FRANCE.POINTS,start=1998,freq=1)
forecast(France.t)

##      Point Forecast     Lo 80    Hi 80     Lo 95    Hi 95
## 2015       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2016       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2017       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2018       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2019       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2020       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2021       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2022       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2023       42.16637 -12.72481 97.05755 -41.78244 126.1152
## 2024       42.16637 -12.72481 97.05755 -41.78244 126.1152

plot(forecast(France.t),xlab="Year",ylab="France Points")

Netherlands

Netherlands.points.1998.2014 <- read.csv("~/R work files/CSV files/Netherlands points 1998-2014.csv")
forecast.Ne <- predict(Netherlands.points.1998.2014$NETHERLANDS.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Netherlands.t <- ts(Netherlands.points.1998.2014$NETHERLANDS.POINTS,start=1998,freq=1)
forecast(Netherlands.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast      Lo 80     Hi 80      Lo 95    Hi 95
## 2010       38.79666   -8.34923  85.94254  -33.30676 110.9001
## 2011       38.79666  -31.52415 109.11746  -68.74974 146.3430
## 2012       38.79666  -57.35849 134.95180 -108.25994 185.8533
## 2013       38.79666  -87.85413 165.44744 -154.89901 232.4923
## 2014       38.79666 -124.83368 202.42699 -211.45434 289.0477
## 2015       38.79666 -170.32945 247.92276 -281.03412 358.6274
## 2016       38.79666 -226.77094 304.36425 -367.35393 444.9472
## 2017       38.79666 -297.14105 374.73436 -474.97573 552.5690
## 2018       38.79666 -385.14440 462.73771 -609.56524 687.1586
## 2019       38.79666 -495.40699 573.00030 -778.19732 855.7906

plot(forecast(Netherlands.t),xlab="Year",ylab="Netherlands Points")

Azerbaijan

Azerbaijan.points.1998.2014 <- read.csv("~/R work files/CSV files/Azerbaijan points 1998-2014.csv")
forecast.Az <- predict(Azerbaijan.points.1998.2014$AZERBAIJAN.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Azerbaijan.t <- ts(Azerbaijan.points.1998.2014$AZERBAIJAN.POINTS,start=1998,freq=1)
forecast(Azerbaijan.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast    Lo 80   Hi 80    Lo 95    Hi 95
## 2005       160.3048 78.32055 242.289 34.92073 285.6888
## 2006       160.3048 78.32055 242.289 34.92073 285.6888
## 2007       160.3048 78.32055 242.289 34.92073 285.6888
## 2008       160.3048 78.32055 242.289 34.92073 285.6888
## 2009       160.3048 78.32055 242.289 34.92073 285.6888
## 2010       160.3048 78.32055 242.289 34.92072 285.6888
## 2011       160.3048 78.32055 242.289 34.92072 285.6888
## 2012       160.3048 78.32055 242.289 34.92072 285.6888
## 2013       160.3048 78.32055 242.289 34.92072 285.6888
## 2014       160.3048 78.32055 242.289 34.92072 285.6888

plot(forecast(Azerbaijan.t),xlab="Year",ylab="Azerbaijan Points")

Russia

Russia.points.1998.2014 <- read.csv("~/R work files/CSV files/Russia points 1998-2014.csv")
forecast.Ru <- predict(Russia.points.1998.2014$RUSSIA.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Russia.t <- ts(Russia.points.1998.2014$RUSSIA.POINTS,start=1998,freq=1)
forecast(Russia.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80     Lo 95    Hi 95
## 2013       136.1424 36.32234 235.9625 -16.51922 288.8041
## 2014       136.1424 36.32234 235.9625 -16.51922 288.8041
## 2015       136.1424 36.32234 235.9625 -16.51922 288.8041
## 2016       136.1424 36.32234 235.9625 -16.51922 288.8041
## 2017       136.1424 36.32234 235.9625 -16.51922 288.8041
## 2018       136.1424 36.32234 235.9625 -16.51923 288.8041
## 2019       136.1424 36.32234 235.9625 -16.51923 288.8041
## 2020       136.1424 36.32234 235.9625 -16.51923 288.8041
## 2021       136.1424 36.32234 235.9625 -16.51923 288.8041
## 2022       136.1424 36.32234 235.9625 -16.51923 288.8041

plot(forecast(Russia.t),xlab="Year",ylab="Russia Points")

Serbia

Serbia.points.1998.2014 <- read.csv("~/R work files/CSV files/Serbia points 1998-2014.csv")
forecast.Se <- predict(Serbia.points.1998.2014$SERBIA.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Serbia.t <- ts(Serbia.points.1998.2014$SERBIA.POINTS,start=1998,freq=1)
forecast(Serbia.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80     Lo 95    Hi 95
## 2006       121.7123 9.818423 233.6062 -49.41462 292.8392
## 2007       121.7123 9.818422 233.6062 -49.41462 292.8392
## 2008       121.7123 9.818422 233.6062 -49.41462 292.8392
## 2009       121.7123 9.818421 233.6062 -49.41462 292.8392
## 2010       121.7123 9.818421 233.6062 -49.41462 292.8392
## 2011       121.7123 9.818420 233.6062 -49.41462 292.8392
## 2012       121.7123 9.818420 233.6062 -49.41462 292.8392
## 2013       121.7123 9.818419 233.6062 -49.41462 292.8392
## 2014       121.7123 9.818419 233.6062 -49.41462 292.8392
## 2015       121.7123 9.818418 233.6062 -49.41463 292.8392

plot(forecast(Serbia.t),xlab="Year",ylab="Serbia Points")

Greece

Greece.points.1998.2014 <- read.csv("~/R work files/CSV files/Greece points 1998-2014.csv")
forecast.G <- predict(Greece.points.1998.2014$GREECE.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Greece.t <- ts(Greece.points.1998.2014$GREECE.POINTS,start=1998,freq=1)
forecast(Greece.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80     Lo 95    Hi 95
## 2012       125.1497 35.22866 215.0708 -12.37268 262.6721
## 2013       125.1497 32.83064 217.4688 -16.04013 266.3396
## 2014       125.1497 30.46353 219.8359 -19.66032 269.9598
## 2015       125.1497 28.12467 222.1748 -23.23729 273.5367
## 2016       125.1497 25.81170 224.4878 -26.77467 277.0741
## 2017       125.1497 23.52250 226.7770 -30.27571 280.5752
## 2018       125.1497 21.25511 229.0443 -33.74338 284.0428
## 2019       125.1497 19.00778 231.2917 -37.18038 287.4798
## 2020       125.1497 16.77890 233.5206 -40.58915 290.8886
## 2021       125.1497 14.56700 235.7325 -43.97196 294.2714

plot(forecast(Greece.t),xlab="Year",ylab="Greece Points")

Armenia

Armenia.points.1998.2014 <- read.csv("~/R work files/CSV files/Armenia points 1998-2014.csv")
forecast.Ar <- predict(Armenia.points.1998.2014$ARMENIA.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Armenia.t <- ts(Armenia.points.1998.2014$ARMENIA.POINTS,start=1998,freq=1)
forecast(Armenia.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80    Lo 95    Hi 95
## 2007       121.0217 53.78226 188.2612 18.18784 223.8556
## 2008       121.0217 53.78226 188.2612 18.18783 223.8556
## 2009       121.0217 53.78226 188.2612 18.18783 223.8556
## 2010       121.0217 53.78226 188.2612 18.18783 223.8556
## 2011       121.0217 53.78226 188.2612 18.18783 223.8556
## 2012       121.0217 53.78226 188.2612 18.18783 223.8556
## 2013       121.0217 53.78226 188.2612 18.18783 223.8556
## 2014       121.0217 53.78226 188.2612 18.18783 223.8556
## 2015       121.0217 53.78226 188.2612 18.18783 223.8556
## 2016       121.0217 53.78226 188.2612 18.18783 223.8556

plot(forecast(Armenia.t),xlab="Year",ylab="Armenia Points")

Italy

Italy.points.1998.2014 <- read.csv("~/R work files/CSV files/Italy points 1998-2014.csv")
forecast.It <- predict(Italy.points.1998.2014$ITALY.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Italy.t <- ts(Italy.points.1998.2014$ITALY.POINTS,start=1998,freq=1)
forecast(Italy.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80    Lo 95    Hi 95
## 2002       112.2566 40.64767 183.8656 2.740184 221.7731
## 2003       112.2566 40.64767 183.8656 2.740183 221.7731
## 2004       112.2566 40.64767 183.8656 2.740182 221.7731
## 2005       112.2566 40.64767 183.8656 2.740182 221.7731
## 2006       112.2566 40.64767 183.8656 2.740181 221.7731
## 2007       112.2566 40.64767 183.8656 2.740180 221.7731
## 2008       112.2566 40.64767 183.8656 2.740180 221.7731
## 2009       112.2566 40.64767 183.8656 2.740179 221.7731
## 2010       112.2566 40.64767 183.8656 2.740178 221.7731
## 2011       112.2566 40.64767 183.8656 2.740178 221.7731

plot(forecast(Italy.t),xlab="Year",ylab="Italy Points")

Denmark

Denmark.points.1998.2014 <- read.csv("~/R work files/CSV files/Denmark points 1998-2014.csv")
forecast.D <- predict(Denmark.points.1998.2014$DENMARK.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Denmark.t <- ts(Denmark.points.1998.2014$DENMARK.POINTS,start=1998,freq=1)
forecast(Denmark.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast     Lo 80    Hi 80     Lo 95   Hi 95
## 2009       90.41887 -5.636575 186.4743 -56.48526 237.323
## 2010       90.41887 -5.636576 186.4743 -56.48526 237.323
## 2011       90.41887 -5.636576 186.4743 -56.48526 237.323
## 2012       90.41887 -5.636577 186.4743 -56.48526 237.323
## 2013       90.41887 -5.636577 186.4743 -56.48526 237.323
## 2014       90.41887 -5.636578 186.4743 -56.48526 237.323
## 2015       90.41887 -5.636578 186.4743 -56.48526 237.323
## 2016       90.41887 -5.636579 186.4743 -56.48526 237.323
## 2017       90.41887 -5.636579 186.4743 -56.48526 237.323
## 2018       90.41887 -5.636580 186.4743 -56.48526 237.323

plot(forecast(Denmark.t),xlab="Year",ylab="Denmark Points")

Norway

Norway.points.1998.2014 <- read.csv("~/R work files/CSV files/Norway points 1998-2014.csv")
forecast.No <- predict(Norway.points.1998.2014$NORWAY.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Norway.t <- ts(Norway.points.1998.2014$NORWAY.POINTS,start=1998,freq=1)
forecast(Norway.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast     Lo 80   Hi 80     Lo 95    Hi 95
## 2010       101.3748 -36.07849 238.828 -108.8419 311.5914
## 2011       101.3748 -36.07849 238.828 -108.8419 311.5914
## 2012       101.3748 -36.07849 238.828 -108.8419 311.5914
## 2013       101.3748 -36.07850 238.828 -108.8419 311.5914
## 2014       101.3748 -36.07850 238.828 -108.8419 311.5914
## 2015       101.3748 -36.07850 238.828 -108.8419 311.5914
## 2016       101.3748 -36.07850 238.828 -108.8419 311.5914
## 2017       101.3748 -36.07850 238.828 -108.8419 311.5914
## 2018       101.3748 -36.07850 238.828 -108.8419 311.5914
## 2019       101.3748 -36.07850 238.828 -108.8419 311.5914

plot(forecast(Norway.t),xlab="Year",ylab="Norway Points")

Romania

Romania.points.1998.2014 <- read.csv("~/R work files/CSV files/Romania points 1998-2014.csv")
forecast.Ro <- predict(Romania.points.1998.2014$ROMANIA.POINTS,h=10)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

Romania.t <- ts(Romania.points.1998.2014$ROMANIA.POINTS,start=1998,freq=1)
forecast(Romania.t)

## Warning in ets(object, lambda = lambda, ...): Missing values encountered.
## Using longest contiguous portion of time series

##      Point Forecast      Lo 80    Hi 80      Lo 95    Hi 95
## 2011         70.298    5.23661 135.3594  -29.20481 169.8008
## 2012         70.298  -20.88407 161.4801  -69.15294 209.7489
## 2013         70.298  -46.93724 187.5332 -108.99783 249.5938
## 2014         70.298  -74.50710 215.1031 -151.16229 291.7583
## 2015         70.298 -104.57468 245.1707 -197.14668 337.7427
## 2016         70.298 -137.95410 278.5501 -248.19609 388.7921
## 2017         70.298 -175.43305 316.0290 -305.51521 446.1112
## 2018         70.298 -217.83671 358.4327 -370.36601 510.9620
## 2019         70.298 -266.06612 406.6621 -444.12653 584.7225
## 2020         70.298 -321.12736 461.7234 -528.33543 668.9314

plot(forecast(Romania.t),xlab="Year",ylab="Romania Points")

Results

The results of my Naïve forecast are as follows: Following the above definition of a Naive forecast, my forecasted winner of the Eurovision song contest 2015 will be Austria due to the fact that they won in the previous year. As an economist, I do not feel this is a very valid forecast, in terms of the nature of this contest. It is very rare that countries win twice in a row; Spain won the 1968 contest, and then drew the following year with UK, Netherlands and France, Luxembourg won both the 1972 and 1973 contests, and finally, Israel won both the 1978 and 1979 contests. These are the only occassions throughout all the 60 years the contest has been running, therefore it would be a wise assumption to state that it is very unlikely that Austria will win the contest again this year.

`2014.full.results` <- read.csv("~/R work files/CSV files/2014 full results.csv")
Full.results.2014 <- `2014.full.results`
library("knitr", lib.loc="~/R/win-library/3.1")
kable(Full.results.2014)

Country	Performer	Song	Place.F	Points.F	Place.SF1	Points.SF1	Place.SF2	Points.SF2	Year
Albania	Hersi	One Night’s Anger	NA	NA	15	22	NA	NA	2014
Armenia	Aram MP3	Not Alone	4	174	4	121	NA	NA	2014
Austria	Conchita Wurst	Rise Like a Phoenix	1	290	NA	NA	1	169	2014
Azerbaijan	Dilara Kazimova	Start A Fire	22	33	9	57	NA	NA	2014
Belarus	Teo	Cheesecake	16	43	NA	NA	5	87	2014
Belgium	Axel Hirsoux	Mother	NA	NA	14	28	NA	NA	2014
Denmark	Basim	Cliche Love Song	9	74	NA	NA	NA	NA	2014
Estonia	Tanja	Amazing	NA	NA	12	36	NA	NA	2014
Finland	Softengine	Something Better	11	72	NA	NA	3	97	2014
France	TWIN TWIN	Moustache	26	2	NA	NA	NA	NA	2014
FYR Macedonia	Tijana	To the Sky	NA	NA	NA	NA	13	33	2014
Georgia	The Shin and Mariko	Three Minutes to Earth	NA	NA	NA	NA	15	15	2014
Germany	Elaiza	Is it right	18	39	NA	NA	NA	NA	2014
Greece	Freaky Fortune feat. RiskyKidd	Rise Up	20	35	NA	NA	7	74	2014
Hungary	András Kállay-Saunders	Running	5	143	3	127	NA	NA	2014
Iceland	Pollapönk	No Prejudice	15	58	8	61	NA	NA	2014
Ireland	Can-Linn (featuring Kasey Smith)	Heartbeat	NA	NA	NA	NA	12	35	2014
Israel	Mei Finegold	Same Heart	NA	NA	NA	NA	14	19	2014
Italy	Emma	La Mia Città	21	33	NA	NA	NA	NA	2014
Latvia	Aarzemnieki	Cake to bake	NA	NA	13	33	NA	NA	2014
Lithuania	Vilija Mata?i?nait?	Attention	NA	NA	NA	NA	11	36	2014
Malta	Firelight	Coming Home	23	32	NA	NA	9	63	2014
Moldova	Cristina Scarlat	Wild Soul	NA	NA	16	13	NA	NA	2014
Montenegro	Sergej ?etkovi?	Moj Svijet	19	37	7	63	NA	NA	2014
Netherlands	The Common Linnets	Calm After The Storm	2	238	1	150	NA	NA	2014
Norway	Carl Espen	Silent Storm	8	88	NA	NA	6	77	2014
Poland	Donatan & Cleo	My S?owianie - We Are Slavic	14	62	NA	NA	8	70	2014
Portugal	Suzy	Quero Ser Tua	NA	NA	11	39	NA	NA	2014
Romania	Paula Seling & OVI	Miracle	12	72	NA	NA	2	125	2014
Russia	Tolmachevy Sisters	Shine	7	89	6	63	NA	NA	2014
San Marino	Valentina Monetta	Maybe (Forse)	24	14	10	40	NA	NA	2014
Slovenia	Tinkara Kova?	Round and round	25	9	NA	NA	10	52	2014
Spain	Ruth Lorenzo	Dancing in the rain	10	74	NA	NA	NA	NA	2014
Sweden	Sanna Nielsen	Undo	3	218	2	131	NA	NA	2014
Switzerland	Sebalter	Hunter Of Stars	13	64	NA	NA	4	92	2014
Ukraine	Mariya Yaremchuk	Tick - Tock	6	113	5	118	NA	NA	2014
United Kingdom	Molly	Children of the Universe	17	40	NA	NA	NA	NA	2014

The results of my Average forecast are as follows: To calculate this forecast, I have chosen to use all of the data from 1998 to 2014. This is due to the fact that telephone voting was introduced in 1998. Therefore throughout this period 1998 to 2014 the voting system has not changed dramatically enough to influence my results. This gives me a very different leaderboard from the one seen above. In this case, my forecasted leaderboard states that Azerbaijan will win the Eurovision song contest 2015. From this we can observe that although Azerbaijan have only entered the contest 7 times, they have been awarded high points each time, including winning the 2011 contest.

Average.forecast <- read.csv("~/R work files/CSV files/Average forecast.csv")
library("knitr", lib.loc="~/R/win-library/3.1")
kable(Average.forecast)

COUNTRY.to.compete.in.2015	X1998	X1999	X2000	X2001	X2002	X2003	X2004	X2005	X2006	X2007	X2008	X2009	X2010	X2011	X2012	X2013	X2014	Total	Average	Place.in.2015
Albania	NA	NA	NA	NA	NA	NA	106.0	53.0	29.0	24.5	55.0	48.0	62.0	23.5	146.0	15.5	11.0	573.5	52.136364	19
Armenia	NA	NA	NA	NA	NA	NA	NA	NA	129.0	138.0	199.0	92.0	141.0	27.0	NA	41.0	174.0	941.0	117.625000	5
Australia	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	0.0	0.000000	40
Austria	NA	65	34	NA	26	101	9.0	15.0	NA	2.0	NA	NA	NA	64.0	4.0	13.5	290.0	623.5	56.681818	18
Azerbaijan	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	132.0	207.0	145.0	221.0	150.0	234.0	33.0	1122.0	160.285714	1
Belarus	NA	NA	NA	NA	NA	NA	5.0	33.5	5.0	145.0	13.5	12.5	18.0	22.5	17.5	48.0	43.0	363.5	33.045455	32
Belgium	122	38	2	NA	33	165	7.0	14.5	34.5	7.0	8.0	0.5	143.0	26.5	8.0	71.0	14.0	694.0	43.375000	26
Cyprus	37	2	8	NA	85	15	170.0	46.0	28.5	32.5	18.0	16.0	27.0	8.0	65.0	5.5	NA	563.5	37.566667	29
Czech Republic	NA	NA	NA	NA	NA	NA	NA	NA	NA	0.5	4.5	0.0	NA	NA	NA	NA	NA	5.0	1.666667	39
Denmark	NA	71	195	177	7	NA	28.0	125.0	26.0	22.5	60.0	74.0	149.0	134.0	21.0	281.0	74.0	1444.5	96.300000	8
Estonia	36	90	98	198	111	14	28.5	15.5	14.0	16.5	4.0	129.0	19.5	44.0	120.0	19.0	18.0	975.0	57.352941	17
Finland	22	NA	18	NA	24	NA	25.5	25.0	292.0	53.0	35.0	22.0	24.5	57.0	20.5	13.0	72.0	703.5	50.250000	21
France	3	14	5	142	104	19	40.0	11.0	5.0	19.0	47.0	107.0	82.0	82.0	21.0	14.0	2.0	717.0	42.176471	27
Georgia	NA	NA	NA	NA	NA	NA	NA	NA	NA	97.0	83.0	NA	136.0	110.0	18.0	50.0	7.5	501.5	71.642857	11
Germany	86	140	96	66	17	53	93.0	4.0	36.0	49.0	14.0	35.0	246.0	107.0	110.0	18.0	39.0	1209.0	71.117647	12
Greece	12	NA	NA	147	27	25	252.0	230.0	128.0	139.0	218.0	120.0	140.0	120.0	64.0	152.0	35.0	1809.0	120.600000	4
Hungary	4	NA	NA	NA	NA	NA	NA	97.0	NA	128.0	3.0	8.0	NA	53.0	19.0	84.0	143.0	539.0	59.888889	16
Iceland	NA	146	45	3	NA	81	16.0	26.0	31.0	38.5	64.0	218.0	41.0	61.0	46.0	47.0	58.0	921.5	61.433333	14
Ireland	64	18	92	6	NA	53	7.0	26.5	93.0	5.0	11.0	26.0	25.0	119.0	46.0	5.0	17.5	614.0	38.375000	28
Israel	172	93	7	25	37	17	28.5	154.0	4.0	8.5	124.0	53.0	71.0	19.0	16.5	20.0	9.5	859.0	50.529412	20
Italy	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	189.0	101.0	126.0	33.0	449.0	112.250000	7
Latvia	NA	NA	136	16	176	5	11.5	153.0	30.0	54.0	83.0	3.5	5.5	12.5	8.5	6.5	16.5	717.5	47.833333	25
Lithuania	NA	13	NA	35	12	NA	13.0	8.5	162.0	28.0	15.0	23.0	22.0	63.0	70.0	17.0	18.0	499.5	35.678571	30
Macedonia (FYR)	16	NA	29	NA	25	NA	47.0	52.0	56.0	73.0	32.0	22.5	18.5	18.0	71.0	14.0	16.5	490.5	35.035714	31
Malta	165	32	73	48	164	4	50.0	192.0	1.0	7.5	19.0	31.0	22.5	27.0	41.0	120.0	32.0	1029.0	60.529412	15
Moldova	NA	NA	NA	NA	NA	NA	NA	148.0	22.0	109.0	18.0	69.0	27.0	97.0	81.0	71.0	6.5	648.5	64.850000	13
Montenegro	NA	NA	NA	NA	NA	NA	NA	NA	NA	16.5	11.5	22.0	NA	NA	10.0	20.5	37.0	117.5	19.583333	37
Netherlands	150	71	40	16	NA	45	11.0	26.5	11.0	19.0	13.5	5.5	14.5	6.5	17.5	114.0	238.0	799.0	49.937500	22
Norway	79	35	57	3	NA	123	3.0	125.0	36.0	24.0	182.0	387.0	35.0	15.0	7.0	191.0	88.0	1390.0	86.875000	9
Poland	19	17	NA	11	NA	90	27.0	40.5	35.0	37.5	14.0	21.5	22.0	9.0	NA	NA	62.0	405.5	31.192308	33
Portugal	36	12	NA	18	NA	13	19.0	25.5	13.0	44.0	69.0	57.0	43.0	11.0	19.5	NA	19.5	399.5	28.535714	35
Romania	6	NA	25	NA	71	73	18.0	158.0	172.0	84.0	45.0	NA	162.0	77.0	71.0	65.0	72.0	1099.0	78.500000	10
Russia	NA	NA	155	37	55	164	67.0	57.0	248.0	207.0	272.0	91.0	90.0	77.0	259.0	174.0	89.0	2042.0	136.133333	2
San Marino	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	2.5	NA	NA	17.0	15.5	23.5	14.0	72.5	14.500000	38
Serbia	NA	NA	NA	NA	NA	NA	NA	NA	NA	268.0	160.0	30.0	72.0	85.0	214.0	23.0	NA	852.0	121.714286	3
Slovenia	17	50	NA	70	33	7	2.5	34.5	24.5	66.0	18.0	7.0	3.0	96.0	15.5	4.0	9.0	457.0	28.562500	34
Spain	21	1	18	76	81	81	87.0	28.0	18.0	43.0	55.0	23.0	68.0	50.0	97.0	8.0	74.0	829.0	48.764706	24
Sweden	53	163	88	100	72	107	170.0	30.0	170.0	51.0	47.0	33.0	31.0	185.0	372.0	62.0	218.0	1952.0	114.823529	6
Switzerland	0	NA	14	NA	15	NA	0.0	128.0	30.0	20.0	23.5	7.5	1.0	19.0	22.5	20.5	32.0	333.0	23.785714	36
UK	166	38	28	28	111	0	29.0	18.0	25.0	19.0	14.0	173.0	10.0	100.0	12.0	23.0	40.0	834.0	49.058824	23

The results from my Simple Exponential Smoothing forecasts and Auto Arima forecasts also follow: N.B. I have used the following countries, as they are the top 5 winning countries, and they are countries from the top 10 on my Average forecasted leaderboard table.

Ireland

ses(Ireland.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast     Lo 80    Hi 80     Lo 95    Hi 95
## 2010       36.31398 -8.341373 80.96933 -31.98049 104.6084
## 2011       36.31398 -8.341373 80.96933 -31.98049 104.6084
## 2012       36.31398 -8.341373 80.96933 -31.98049 104.6084
## 2013       36.31398 -8.341373 80.96933 -31.98049 104.6084
## 2014       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2015       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2016       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2017       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2018       36.31398 -8.341374 80.96933 -31.98049 104.6084
## 2019       36.31398 -8.341375 80.96933 -31.98049 104.6084

auto.arima(Ireland.t)

## Series: Ireland.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##          38.375
## s.e.      8.755
## 
## sigma^2 estimated as 1226:  log likelihood=-79.6
## AIC=163.2   AICc=164.05   BIC=164.86

summary(Ireland.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    5.00   10.00   25.50   38.38   55.75  119.00       1

UK

ses(uk.t)

##      Point Forecast     Lo 80    Hi 80     Lo 95    Hi 95
## 2015        49.3248 -17.79317 116.4428 -53.32327 151.9729
## 2016        49.3248 -17.79317 116.4428 -53.32327 151.9729
## 2017        49.3248 -17.79317 116.4428 -53.32327 151.9729
## 2018        49.3248 -17.79317 116.4428 -53.32327 151.9729
## 2019        49.3248 -17.79317 116.4428 -53.32327 151.9729
## 2020        49.3248 -17.79317 116.4428 -53.32327 151.9729
## 2021        49.3248 -17.79317 116.4428 -53.32327 151.9729
## 2022        49.3248 -17.79317 116.4428 -53.32328 151.9729
## 2023        49.3248 -17.79317 116.4428 -53.32328 151.9729
## 2024        49.3248 -17.79317 116.4428 -53.32328 151.9729

auto.arima(uk.t)

## Series: uk.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##         49.0588
## s.e.    12.7014
## 
## sigma^2 estimated as 2743:  log likelihood=-91.41
## AIC=186.83   AICc=187.68   BIC=188.49

summary(uk.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    0.00   18.00   28.00   49.06   40.00  173.00

Sweden

ses(Sweden.t)

##      Point Forecast    Lo 80    Hi 80     Lo 95  Hi 95
## 2015       114.8445 3.134697 226.5544 -56.00092 285.69
## 2016       114.8445 3.134696 226.5544 -56.00092 285.69
## 2017       114.8445 3.134696 226.5544 -56.00092 285.69
## 2018       114.8445 3.134695 226.5544 -56.00092 285.69
## 2019       114.8445 3.134695 226.5544 -56.00092 285.69
## 2020       114.8445 3.134694 226.5544 -56.00092 285.69
## 2021       114.8445 3.134694 226.5544 -56.00093 285.69
## 2022       114.8445 3.134693 226.5544 -56.00093 285.69
## 2023       114.8445 3.134692 226.5544 -56.00093 285.69
## 2024       114.8445 3.134692 226.5544 -56.00093 285.69

auto.arima(Sweden.t)

## Series: Sweden.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##        114.8235
## s.e.    21.1403
## 
## sigma^2 estimated as 7597:  log likelihood=-100.07
## AIC=204.15   AICc=205.01   BIC=205.81

summary(Sweden.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    30.0    51.0    88.0   114.8   170.0   372.0

France

ses(France.t)

##      Point Forecast     Lo 80    Hi 80     Lo 95    Hi 95
## 2015       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2016       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2017       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2018       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2019       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2020       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2021       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2022       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2023       42.17554 -12.71564 97.06671 -41.77327 126.1243
## 2024       42.17554 -12.71564 97.06671 -41.77327 126.1243

auto.arima(France.t)

## Series: France.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##         42.1765
## s.e.    10.3878
## 
## sigma^2 estimated as 1834:  log likelihood=-87.99
## AIC=179.99   AICc=180.85   BIC=181.66

summary(France.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    2.00   11.00   19.00   42.18   82.00  142.00

Netherlands

ses(Netherlands.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast     Lo 80    Hi 80     Lo 95    Hi 95
## 2010       43.50572 -40.07365 127.0851 -84.31789 171.3293
## 2011       43.50572 -40.07365 127.0851 -84.31789 171.3293
## 2012       43.50572 -40.07365 127.0851 -84.31789 171.3293
## 2013       43.50572 -40.07365 127.0851 -84.31790 171.3293
## 2014       43.50572 -40.07365 127.0851 -84.31790 171.3293
## 2015       43.50572 -40.07365 127.0851 -84.31790 171.3293
## 2016       43.50572 -40.07365 127.0851 -84.31790 171.3293
## 2017       43.50572 -40.07365 127.0851 -84.31790 171.3293
## 2018       43.50572 -40.07365 127.0851 -84.31790 171.3293
## 2019       43.50572 -40.07365 127.0851 -84.31790 171.3293

auto.arima(Netherlands.t)

## Series: Netherlands.t 
## ARIMA(2,0,0) with zero mean     
## 
## Coefficients:
##          ar1      ar2
##       1.6761  -0.7354
## s.e.  0.2213   0.2156
## 
## sigma^2 estimated as 1388:  log likelihood=-83.45
## AIC=172.9   AICc=174.75   BIC=175.4

summary(Netherlands.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    5.50   12.88   18.25   49.94   51.50  238.00       1

Azerbaijan

ses(Azerbaijan.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80    Lo 95    Hi 95
## 2005        160.652 78.66816 242.6359 35.26853 286.0355
## 2006        160.652 78.66816 242.6359 35.26853 286.0355
## 2007        160.652 78.66816 242.6359 35.26852 286.0355
## 2008        160.652 78.66816 242.6359 35.26852 286.0355
## 2009        160.652 78.66816 242.6359 35.26852 286.0355
## 2010        160.652 78.66816 242.6359 35.26852 286.0355
## 2011        160.652 78.66816 242.6359 35.26852 286.0355
## 2012        160.652 78.66816 242.6359 35.26852 286.0355
## 2013        160.652 78.66816 242.6359 35.26852 286.0355
## 2014        160.652 78.66816 242.6359 35.26852 286.0355

auto.arima(Azerbaijan.t)

## Series: Azerbaijan.t 
## ARIMA(1,0,0) with non-zero mean 
## 
## Coefficients:
##           ar1  intercept
##       -0.9021   172.4778
## s.e.   0.1408     8.1977
## 
## sigma^2 estimated as 1442:  log likelihood=-36.23
## AIC=78.46   AICc=80.31   BIC=80.96

summary(Azerbaijan.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    33.0   138.5   150.0   160.3   214.0   234.0      10

Russia

ses(Russia.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80     Lo 95    Hi 95
## 2013       136.1318 36.31172 235.9519 -16.52984 288.7934
## 2014       136.1318 36.31172 235.9519 -16.52984 288.7934
## 2015       136.1318 36.31172 235.9519 -16.52984 288.7934
## 2016       136.1318 36.31172 235.9519 -16.52984 288.7934
## 2017       136.1318 36.31172 235.9519 -16.52984 288.7934
## 2018       136.1318 36.31172 235.9519 -16.52984 288.7934
## 2019       136.1318 36.31172 235.9519 -16.52984 288.7934
## 2020       136.1318 36.31171 235.9519 -16.52984 288.7934
## 2021       136.1318 36.31171 235.9519 -16.52985 288.7934
## 2022       136.1318 36.31171 235.9519 -16.52985 288.7934

auto.arima(Russia.t)

## Series: Russia.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##        136.1333
## s.e.    20.1102
## 
## sigma^2 estimated as 6066:  log likelihood=-86.61
## AIC=177.23   AICc=178.08   BIC=178.89

summary(Russia.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    37.0    72.0    91.0   136.1   190.5   272.0       2

Serbia

ses(Serbia.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80     Lo 95    Hi 95
## 2006       121.7207 9.826816 233.6145 -49.40621 292.8475
## 2007       121.7207 9.826816 233.6145 -49.40621 292.8475
## 2008       121.7207 9.826815 233.6145 -49.40621 292.8475
## 2009       121.7207 9.826814 233.6145 -49.40621 292.8476
## 2010       121.7207 9.826814 233.6145 -49.40621 292.8476
## 2011       121.7207 9.826813 233.6145 -49.40622 292.8476
## 2012       121.7207 9.826813 233.6145 -49.40622 292.8476
## 2013       121.7207 9.826812 233.6145 -49.40622 292.8476
## 2014       121.7207 9.826812 233.6145 -49.40622 292.8476
## 2015       121.7207 9.826811 233.6145 -49.40622 292.8476

auto.arima(Serbia.t)

## Series: Serbia.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##        121.7143
## s.e.    32.9992
## 
## sigma^2 estimated as 7622:  log likelihood=-41.22
## AIC=86.44   AICc=87.29   BIC=88.1

summary(Serbia.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    23.0    51.0    85.0   121.7   187.0   268.0      10

Greece

ses(Greece.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80     Lo 95    Hi 95
## 2012       133.3522 43.30429 223.4001 -4.364201 271.0686
## 2013       133.3522 43.30429 223.4001 -4.364202 271.0686
## 2014       133.3522 43.30429 223.4001 -4.364203 271.0686
## 2015       133.3522 43.30429 223.4001 -4.364204 271.0686
## 2016       133.3522 43.30429 223.4001 -4.364204 271.0686
## 2017       133.3522 43.30429 223.4001 -4.364205 271.0686
## 2018       133.3522 43.30429 223.4001 -4.364206 271.0686
## 2019       133.3522 43.30429 223.4001 -4.364206 271.0686
## 2020       133.3522 43.30429 223.4001 -4.364207 271.0686
## 2021       133.3522 43.30429 223.4001 -4.364208 271.0686

auto.arima(Greece.t)

## Series: Greece.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##        120.6000
## s.e.    19.0205
## 
## sigma^2 estimated as 5427:  log likelihood=-85.78
## AIC=175.55   AICc=176.41   BIC=177.22

summary(Greece.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    12.0    49.5   128.0   120.6   149.5   252.0       2

Armenia

ses(Armenia.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80    Lo 95    Hi 95
## 2007        120.961 53.72402 188.1981 18.13090 223.7912
## 2008        120.961 53.72402 188.1981 18.13090 223.7912
## 2009        120.961 53.72402 188.1981 18.13090 223.7912
## 2010        120.961 53.72402 188.1981 18.13089 223.7912
## 2011        120.961 53.72402 188.1981 18.13089 223.7912
## 2012        120.961 53.72402 188.1981 18.13089 223.7912
## 2013        120.961 53.72402 188.1981 18.13089 223.7912
## 2014        120.961 53.72402 188.1981 18.13089 223.7912
## 2015        120.961 53.72402 188.1981 18.13089 223.7912
## 2016        120.961 53.72402 188.1981 18.13089 223.7912

auto.arima(Armenia.t)

## Series: Armenia.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##        117.6250
## s.e.    20.0126
## 
## sigma^2 estimated as 3204:  log likelihood=-43.64
## AIC=91.28   AICc=92.14   BIC=92.95

summary(Armenia.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##   27.00   79.25  133.50  117.60  149.20  199.00       9

Italy

ses(Italy.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80    Lo 95   Hi 95
## 2002       112.2493 40.64278 183.8558 2.736578 221.762
## 2003       112.2493 40.64278 183.8558 2.736578 221.762
## 2004       112.2493 40.64277 183.8558 2.736577 221.762
## 2005       112.2493 40.64277 183.8558 2.736577 221.762
## 2006       112.2493 40.64277 183.8558 2.736576 221.762
## 2007       112.2493 40.64277 183.8558 2.736576 221.762
## 2008       112.2493 40.64277 183.8558 2.736575 221.762
## 2009       112.2493 40.64277 183.8558 2.736574 221.762
## 2010       112.2493 40.64277 183.8558 2.736574 221.762
## 2011       112.2493 40.64277 183.8558 2.736573 221.762

auto.arima(Italy.t)

## Series: Italy.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##        112.2500
## s.e.    27.9365
## 
## sigma^2 estimated as 3122:  log likelihood=-21.77
## AIC=47.54   AICc=48.39   BIC=49.2

summary(Italy.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    33.0    84.0   113.5   112.2   141.8   189.0      13

Denmark

ses(Denmark.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast     Lo 80    Hi 80     Lo 95   Hi 95
## 2009       90.40991 -5.645539 186.4654 -56.49422 237.314
## 2010       90.40991 -5.645540 186.4654 -56.49422 237.314
## 2011       90.40991 -5.645540 186.4654 -56.49422 237.314
## 2012       90.40991 -5.645541 186.4654 -56.49422 237.314
## 2013       90.40991 -5.645541 186.4654 -56.49422 237.314
## 2014       90.40991 -5.645542 186.4654 -56.49422 237.314
## 2015       90.40991 -5.645542 186.4654 -56.49422 237.314
## 2016       90.40991 -5.645543 186.4654 -56.49423 237.314
## 2017       90.40991 -5.645543 186.4654 -56.49423 237.314
## 2018       90.40991 -5.645544 186.4654 -56.49423 237.314

auto.arima(Denmark.t)

## Series: Denmark.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##         96.3000
## s.e.    19.6685
## 
## sigma^2 estimated as 5803:  log likelihood=-86.28
## AIC=176.56   AICc=177.42   BIC=178.23

summary(Denmark.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##     7.0    27.0    74.0    96.3   141.5   281.0       2

Norway

ses(Norway.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast     Lo 80   Hi 80     Lo 95    Hi 95
## 2010       101.3698 -36.08348 238.823 -108.8468 311.5864
## 2011       101.3698 -36.08348 238.823 -108.8468 311.5864
## 2012       101.3698 -36.08348 238.823 -108.8468 311.5864
## 2013       101.3698 -36.08348 238.823 -108.8468 311.5864
## 2014       101.3698 -36.08348 238.823 -108.8468 311.5864
## 2015       101.3698 -36.08348 238.823 -108.8468 311.5864
## 2016       101.3698 -36.08349 238.823 -108.8468 311.5864
## 2017       101.3698 -36.08349 238.823 -108.8468 311.5864
## 2018       101.3698 -36.08349 238.823 -108.8468 311.5864
## 2019       101.3698 -36.08349 238.823 -108.8468 311.5864

auto.arima(Norway.t)

## Series: Norway.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##         86.8750
## s.e.    24.3044
## 
## sigma^2 estimated as 9451:  log likelihood=-95.93
## AIC=195.87   AICc=196.73   BIC=197.53

summary(Norway.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    3.00   21.75   46.50   86.88  123.50  387.00       1

Romania

ses(Romania.t)

## Warning in ets(x, "ANN", alpha = alpha, opt.crit = "mse"): Missing values
## encountered. Using longest contiguous portion of time series

##      Point Forecast    Lo 80    Hi 80     Lo 95    Hi 95
## 2011       88.10187 20.92122 155.2825 -14.64207 190.8458
## 2012       88.10187 20.92122 155.2825 -14.64207 190.8458
## 2013       88.10187 20.92122 155.2825 -14.64207 190.8458
## 2014       88.10187 20.92122 155.2825 -14.64207 190.8458
## 2015       88.10187 20.92122 155.2825 -14.64207 190.8458
## 2016       88.10187 20.92121 155.2825 -14.64208 190.8458
## 2017       88.10187 20.92121 155.2825 -14.64208 190.8458
## 2018       88.10187 20.92121 155.2825 -14.64208 190.8458
## 2019       88.10187 20.92121 155.2825 -14.64208 190.8458
## 2020       88.10187 20.92121 155.2825 -14.64208 190.8458

auto.arima(Romania.t)

## Series: Romania.t 
## ARIMA(0,0,0) with non-zero mean 
## 
## Coefficients:
##       intercept
##          78.500
## s.e.     13.427
## 
## sigma^2 estimated as 2524:  log likelihood=-74.7
## AIC=153.4   AICc=154.26   BIC=155.07

summary(Romania.t)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    6.00   50.00   71.50   78.50   82.25  172.00       3

Conclusion

In all of my models and forecasts I have found that the winning country of the Eurovision song contests 2015 is to be Azerbaijan (except for the result of my Naïve forecast which gave me Austria as the winning country). Above I have explained why Austria winning the contest this year is unlikely, therefore I am inclined to trust my SES forecast. Azerbaijan last won in 2011, therefore it is potentially a possibility that they win this year. You do not need to be an economist to see that the above ETS graphs are not very precise. The blue forecasted areas are very wide, and therefore I cannot rely on these.

In the following table I have attached the announced entries for this years ESC and languages the songs will be performed in. Where the song title is in another language, I have also added a column holding the translation. the table shows that Azerbaijan plan to sing their entry in English. It is necessary to remember that I have a lack of data for Australia, therefore they have not been included in any of my forecasts.

Entries.for.2015 <- read.csv("~/R work files/CSV files/Entries for 2015.csv")
library("knitr", lib.loc="~/R/win-library/3.1")
kable(Entries.for.2015)

COUNTRY.to.compete.in.2015	Song.title	Translation	Language
Albania	I’m alive		English
Armenia	Face the shadow		English
Australia	Tonight again		English
Austria	I am yours		English
Azerbaijan	Hour of the wolf		English
Belarus	Time		English
Belgium	Rhythm inside		English
Cyprus	One thing I should have done		English
Czech Republic	Hope never dies		English
Denmark	The way you are		English
Estonia	Goodbye to yesterday		English
Finland	Aina Mun Pitää	I always have to	Finnish
France	Noubliez pas	Don’t forget	French
Georgia	Warrior		English
Germany	Black smoke		English
Greece	One last breath		English
Hungary	Wars for nothing		English
Iceland	Unbroken		English
Ireland	Playing with numbers		English
Israel	Golden boy		English
Italy	Grande Amore	Great Love	Italian
Latvia	Love injected		English
Lithuania	This time		English
Macedonia	Autumn leaves		English
Malta	Warrior		English
Moldova	I want your love		English
Montenegro	Adio	Goodbye	Croatian
Netherlands	Walk along		English
Norway	A monster like me		English
Poland	In the name of love		English
Portugal	Há um mar que nos Separa	There’s a sea that separates us	Portugese
Romania	De la cap?t	All over again	Romanian
Russia	A million voices		English
San Marino	Chain of light		English
Serbia	Beauty never lies		English
Slovenia	Here for you		English
Spain	Amanecer	Dawn	Spanish
Sweden	Heroes		English
Switzerland	Time to shine		English
UK	Still in love		English

Bookies favourite however is Sweden with a price of 2/1, and they are currently holding Azerbaijan in 9th at an average of 20/1.

It is also worth noting that there is a lot of suspicion about voting bias. Although countries are supposed to vote based on the quality of the entry, it is common for countries to vote based on geographical location, for example, it is known that the following are very strong friendships: - Andorra and Spain - Romania and Moldova - Greece and Cyprus - Turkey and Azerbaijan - Georgia and Lithuania - Montenegro and Albania - Albania and Macedonia - Macedonia and Serbia - San Marino and Albania - Montenegro and Macedonia

semi.final.pots <- read.csv("~/R work files/CSV files/semi final pots.csv")
kable(semi.final.pots)

Pot.1	Pot.2	Pot.3	Pot.4	Pot.5	Guaranteed.Finalists
Albania	Denmark	Armenia	Belgium	Czech Republic	Austria
Macedonia	Estonia	Azerbaijan	Cyprus	Hungary	France
Malta	Finland	Belarus	Greece	Moldova	Germany
Montenegro	Iceland	Georgia	Ireland	Poland	Italy
Serbia	Latvia	Israel	Netherlands	Portugal	Spain
Slovenia	Norway	Lithuania	San Marino	Romania	UK
Switzerland	Sweden	Russia			Australia

Above are the semi-final pots in which the countries will be competing for a place in the final. The column on the far right shows the Guaranteed finalists, which include the host nation, the ‘Big Five’ sponsors and Australia (making their debut).