IRJET- Weather Prediction for Tourism Application using ARIMA

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 11 | Nov 2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 3586
Weather Prediction for Tourism Application using ARIMA
Abhijit Kocharekar1, Bharat V. Nemade2, Chetan G. Patil3, Durgesh D. Sapkale4,
Prof. Sagar G. Salunke5
1,2,3,4Computer Engineering Department, Pimpri Chinchwad College of Engineering, Pune, India
5Project Guide, Computer Engineering Department, Pimpri Chinchwad College of Engineering, Pune, India
----------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In many areas, accurate projections of future
occurrences are crucial, one of which is the tourism industry.
Usually counter-trials and towns spend a enormous quantity
of cash in planning and preparation to accommodate (and
benefit) visitors. Precisely predicting the amount of visits in
the days or months that follow could assist both the economy
and tourists.
Previous studies in this field investigate predictions for a
nation as a whole rather than for fine-grained fields within a
nation. Weather forecasting has drawn the attention of many
scientists from distinct research communitiesduetoitsimpact
on human life globally. The developing deep learning methods
coupled with the wide accessibility of huge weather
observation data and the advent of informationandcomputer
technology have motivated many scientists to investigate
hidden hierarchical patterns for weather forecasting in large
amounts of weather data over the previous century.
To predict climate information accurately, heavy statistical
algorithms are used on the big quantity of historical
information. Time series Analysis enables us know the
fundamental forces leading to a specific trend in time series
data points and enables us to predict andmonitorinformation
points by fitting suitable models into them.
In this study, ARIMA model is used for predicting time series.
ARIMA is an acronym representing the AutoRegressive
Integrated Moving Average. It is a model classthatcapturesin
time series data a suite of distinct normal temporal stuctures.
Key Words: Tourism Industry, Weather Forecasting,
Time Series Analysis, ARIMA
1. INTRODUCTION
Climate and weather are essential considerations in the
decision making for visitors and also affect the effective
operation of tourism enterprises. Tourist industry is a
contributing sector to the global economy. Indeed, the
economies of some nations derive most oftheirincomefrom
tourism. The rise in individual revenue andthepromotionof
their attractions by distinct nations led the sector to evolve.
For the economy of the country, tourism in India is essential
and is increasing quickly. The World Travel and Tourism
Council calculated that tourism generated 16.91 lakh crore
or 9.2% of India's GDP in 2018 and provided 42.673 million
employment, 8.1% of total employment. The industry is
forecast to expand by 2028 (9.9 % of GDP) at an annual rate
of 6.9% to 32.05 lakh crore. All tourist destinations are
climate sensitive and climate has a main impact on travel
planning and travel experience. Many kinds of tourism
depend on the weather and, by extension, depend on the
climate. Therefore, sooner or later, climate change is
probable to impact your business area. Climate change can,
for instance, decrease snow cover, boost and prolong heat
waves, or change annual rainfall patterns.Risk identification
can be accomplished by studying this climate change and its
effect on the tourism industry. Proper tourism management
and tour planning can be efficiently done byevaluatingthese
risk variables. Hence proper measures can be taken by the
government and holiday planners can effectively plan the
tours. Weather forecasting is an appealing research topic
with extensive potential applications ranging from flight
navigation to farming and tourism. Also other thrust areas
where weather forecasting can be proved to be essential
include Air Traffic Control (ATC), Voyage planning, Military
applications, Transport industry etc. Weather forecasting
can also have a significant effect on various sports.
Intelligent systems based on machine learning algorithms
have the ability to learn from previous knowledge or
historical information and thus have received important
recognition in the Computer Science Community. Weather
Prediction and Forecasting is an application of science,
research and technology to predicttheclimatefora specified
place and specified instance of time using machine learning
algorithms. The weather forecasting problems, among
others, are learning weather representation using a huge
weather dataset quantity. Analysis of various information
mining procedures is carried out for this purpose. Data
mining methods allow users to analyze, classify and
condense the known associationsfroma broad rangeofsizes
or angles. Classification, learning and prediction are some
basic terms linked to data mining.
2. RELATED WORKS
Related works included many distinct and exciting weather
forecasting methods. While much of the present prediction
technology includes physics-based simulations and
differential equations, many fresh methods from artificial
intelligence primarily used machine learning methods,
mostly neural networks, while some used probabilistic
models such as Bayesian networks. Two of the three papers
on weather prediction machine learning we analyzed used
neural networks while one used support vector machines.
Neural networks, unlike the linear regressionandfunctional

regression models we used,seemtobethecommonmachine
learning model option for weather forecasting due to the
capacity to capture the non-linear dependencies of previous
weather trends and future weatherconditions.Thisgives the
benefit of not assuming that all characteristics are easy
linear dependencies over our models.Ofthetwosolutionsto
neural networks, one used a hybrid model that used neural
networks to model physicsbehindweatherforecastingwhile
the other[1] used learning more directly to predict weather
conditions. Similarly, the strategy using support vector
machines[2] also directly implemented the classifier for
weather prediction but was more restricted in scope than
approaches to the neural network.
Over the last decade, countless important attempts have
been documented with effective outcomes to fix weather
forecasting issues using statistical modeling, including
machine learning systems[3][4 ].Different methods have
been used in weather prediction systems such as neural
network-based algorithms using Back Propagation Neural
Network (BPN) and Hopfield Network[5 ], predictive
analysis in Apache Hadoop Framework utilizingNaiveBayes
Algorithm [7], Artificial Neural Network and Decision Tree
Algorithms[6 ],Recurrence Neural Network (RNN),
Conditional Restricted Boltzmann Machine (CRBM) and
Convolutional Network (CN) models[8 ].
A few studies concentrate on evaluating information from
social media, such as check-ins and remarks published by
visitors, to infer the density of visitation over time. For
example, in (Spencer A. Wood and Lacayo 2013)[9], the
writers use the locations of photos in Flickr, a popular
website for picture hosting, to estimate the number of
visitors to some recreational sites around the globe. They
study the link between the empirical estimates of mean
annual user-days visitors and photo-derived ones. This is
best described by a polynomial function with R2=0.386and
that categorizing the recreational parts into more specific
profiles could improve correlations. However, they do not
address predictions. In (Fisichelli et al. 2015)[10], the
writers use a third-order polynomial temperature model to
evaluate the climate and visitation information for U.S.
national parks and claim that it explains 69 percent of the
variability in historical visitation patterns. We were able to
attain much greater levels of precision by exploiting a richer
function set, including social media information. We
demonstrate that even a simple linear regression model can
achieve an precision of over 80 percent by exploiting social
media as well as climate information, while a more robust
algorithm, Support Vector Regression, generated a superior
94 percent precision outcome. The writers use Wikipedia
usage trends in more latest work (Khadivi and
Ramakrishnan 2016)[11] to predictHawaii'stravel demand.
However, they only report the precision of their forecast
outcomes by using an auto-regressive exogenous model
where the external variable is a trend time series of
Wikipedia usage. RMSE is an accuracy metric to compare
forecast mistakes of distinct models for specific information
and not between datasets as it is scale dependent(Hyndman
and Koehler 2006)[12]. Although this work contains
interesting statementsandoutcomes,thereisnocomparison
of predictive models with other baselines or similar
evaluation of the outcomes.
3. PROPOSED METHOD
Fig -1: Workflow of the proposed system
Data Acquisition:
The weather data set supplied by the Indian Meteorological
Department has been taken into account, consisting of
various features such as air temperature, atmospheric
pressure, relative humidity, average wind speed, maximum
air temperature, complete cloud cover, horizontal visibility,
minimum dew point air temperature.Thedatasetcontainsa
total of eight records for one day. Each of these is held after
every 3 hours of day. The dataset includes a total of 23,000
records.
Data Preprocessing:
As part of pre-processing, the missing values must be
removed from the dataset to ensure that the results
generated are more accurate. Since most data is available in
numerical format, it is possible to calculate the mean,
median or mode of the features and replace it with missing
values. This is an approximation that can add variancetothe
dataset.Replacing the three approximations above is a
statistical approach to the handling of missing values.After
there are eight readings for each day, it must be normalised
in such a way that for the particular day only one reading is
obtained.

Data visualization:
Data visualization provides additional interpretation possibil
ities.Visual presentation for individuals is often more readab
le than any other presentation of dataWeather data in itself i
s massive. Due to the fact that the data is in enormous amoun
ts, data visualization helps to understand the dataset. It helps
to see patterns, trends and correlation in the attribute
values that may go unnoticed. It also helps to identify climate
data patterns and their variations throughout the year
Data Conversion
Weather data may be in non-stationary time series with
statistical properties that change overtime.Itisnecessaryto
ensure that these statistical properties are constant before
starting any predictive modelling. A stationary time series is
one whose statistical properties are relatively constantover
time, such as mean, variance and auto correlation.
Conversion to stationary time series using:
a. Differencing: In differencing, the difference between the
consecutive terms is calculated. Differencing is typically
performed to get rid of the varying mean. Making a
Stationary time series through differencing is an important
part of the process of fitting an ARIMA model.
Mathematically, differencing can be written as:
yt‘=yt – y(t-1) ,where yt is the value at a time t.
b. Transformation: To stabilize a series' non-constant
variance, transformations are used. Common methods of
transformation include power transformation, square root
transformation and log transformation.
Model analysis
Weather forecasting is the system's main objective. ARIMA
model is the most suitable statistical method for this
purpose. ARIMA model: ARIMAisanacronymthatstandsfor
Auto-Regressive Integrated Moving Average. An ARIMA
model is a class of statistical models for time series data
analysis and forecasting. The acronym is descriptive,
a. AR: Autoregression.A model that uses an observation's
dependent relationship with a number of lagged
observations
b. I: Integrated.The use of differencing ofrawobservationsin
order to make the time series stationary.
c. MA: Moving Average. A model that uses the dependency
between an observation and a residual error from a moving
average model applied to lagged observations. Each of these
components is specified explicitly as a parameter in the
model.
4. ALGORITHM
A time series is a set of observations on the values taken at
different times by a variable.
Such information must be gathered daily (e.g. weather),
weekly (e.g. monthly supply) or annually (e.g. government
budget) at periodic intervals.
Time Series is used in statistics, finance, prediction of
earthquake, forecasting of weather and many other
applications.
Stationary Series:
There are three basic criterion for a series to be classified as
stationary series:
1. The mean of the series should not be a function of time
rather should be a constant. The image below has the left
hand graph satisfying the conditionwhereasthegraphin red
has a time dependent mean.
Fig -2: Time independent mean vs Time independent
mean
2. The variance of the series should not a be a function of
time. This property is known as homoscedasticity.
Following graph depicts what is and what is not a stationary
series.
The varying spread of distribution in the right hand graph,
indicates it is non-stationary.
Fig -3: Spread of Distribution in Stationary and Non-
Stationary series

3. The covariance of the i th term and the (i + m) th term
should not be a function of time.
In the following graph, you will notice the spread becomes
closer as the time increases.
Hence, the covariance is not constant with time for the ‘red
series’.
Fig -4: Covariance in Stationary and Non- Stationary series
Test for Stationarity of Series:
Plotting Rolling Statistics:
In this method ,the moving average or moving variance and
see if it varies with time.
By moving average or variance ,it means that at any instant
‘t’, we’ll take the average or variance of the last year, i.e. last
12 months.
But again Rolling Statistics is more of a visual technique.
Dickey-Fuller Test:
This is one of the statistical tests for checking stationarity.
Here the null hypothesis is that the Time Series is non-
stationary.
The test results comprise of a Test Statistic andsomeCritical
Values for difference confidence levels.
If the ‘Test Statistic’ is less than the ‘Critical Value’, we can
reject the null hypothesis and say that the series is
stationary.
ARIMA MODEL
ARIMA stands for Auto-Regressive Integrated Moving
Averages.
The ARIMA forecasting for a stationarytimeseriesisnothing
but a linear (like a linear regression) equation.
Number of AR (Auto-Regressive) terms (p): AR terms are
just lags of dependent variable. For instance if p is 5, the
predictors for x(t) will be x(t-1)….x(t-5).
Number of MA (Moving Average) terms(q): MAtermsare
lagged forecast errors in prediction equation. For instanceif
q is 5, the predictors for x(t) will be e(t-1)….e(t-5)where e(i)
is the difference between the moving average at ith instant
and actual value.
Number of Differences (d): These are the number of non-
seasonal differences, i.e. in most case we took the first order
difference. For d=2,it means that the differences has been
calculated 2 times. By visualizing these newlycreatedseries,
we can identify ideal transformation for our use-case.
5. CONCLUSION
The primary goal of this project is that, the tourist should be
able to plan his holidays/trips based on the predictions
generated by the proposed system .This System should be
able to predict the suitable weather forecast so that the
tourist can have an a reliable application where he can
simply enter the date and duration of his tour and hence
validate it againstthepredictionsanddecisionsgenerated by
the System. Compared to traditional machine learning
models such as Regression and Classification, Time series
particularly, ARIMA model can deliver higher accuracies for
the prediction. Since, this idea is not implementedassuchby
various Big Tourist Industry giants, it can be a new
development area to be explored which can definitely
benefit the tourist by the predictions generated by our
system.
REFERENCES
[1] Lai, Loi Lei, et al. ”Intelligent weather forecast.”Machine
Learning and Cybernetics, 2004. Proceedings of 2004
International Conference on. Vol. 7. IEEE, 2004.
[2] Radhika, Y., and M. Shashi. ”Atmospheric temperature
prediction usingsupportvectormachines.”International
Journal of Computer Theory and Engineering1.1
(2009):55.
[3] A. G. Salman, B. Kanigoro and Y. Heryadi, "Weather
forecasting using deep learning techniques," 2015
International Conference on Advanced Computer
Science and Information Systems (ICACSIS), 2015,
pp.281-285.
[4] Chen, S.-M., and J.-R. Hwang. "Temperature prediction
using fuzzy time series." Systems, Man, and Cybernetics,
Part B: Cybernetics, IEEE Transactions on 30.2
(2000):263-275.
[5] Ghosh et al., "Weather Data Mining using Artificial
Neural Network," 2011 IEEE Recent Advances in

Intelligent Computational Systems, Trivandrum, 2011,
pp. 192-195.
[6] Wang, ZhanJie & Mujib, A B M. (2017). “The Weather
Forecast Using Data Mining Research Based on Cloud
Computing”. Journal of Physics, pp 1-6.
[7] Mr. Sunil Navadia, Mr. Jobin Thomas, Mr. Pintukumar
Yadav, Ms. Shakila Shaikh, “Weather Prediction: Anovel
approach for measuring and analyzing weather data”,
International conference on I-SMAC (IoT in Social,
Mobile, Analytics and Cloud), (I-SMAC 2017), IEEE, pp
414-417
[8] A. G. Salman, B. Kanigoro and Y. Heryadi, "Weather
forecasting using deep learning techniques," 2015
International Conference on Advanced Computer
Science and Information Systems (ICACSIS), 2015,
pp.281-285.
[9] Spencer A. Wood, Anne D. Guerry, J. M. S., and Lacayo,M.
2013. Using social media to quantify nature-based
tourism and recreation. Scientific Report 3.
[10] Fisichelli, N. A.; Schuurman, G. W.; Monahan, W. B.; and
Ziesler, P. S. 2015. Protected area tourism in a changing
climate: Will visitation at us national parks warm up or
overheat? PLoS ONE 10(6).
[11] Khadivi, P., and Ramakrishnan,N.2016.Wikipedia inthe
tourism industry: Forecasting demand and modeling
usage behavior. In ICWSM, 4016–4021.
[12] Hyndman, R. J., and Koehler, A. B. 2006. Another look at
measures o forecast accuracy. Int. J. Forecasting
22(4):679–688.
[13] Stern, H. (2008), The accuracy of weather forecasts for
Melbourne, Australia. Met. Apps, 15: 65?71.
doi:10.1002/met.67
[14] Abramson, Bruce, et al. ”Hailfinder: A Bayesian system
for forecasting severe weather.”International Journal of
Forecasting12.1 (1996): 57-71.
[15] Paras, S. Mathur, A. Kumar and M. Chandra (2007), "A
Feature Based Neural Network Model for Weather
Forecasting", World Academy of Science, Engineering
and Technology.
[16] Afan Galih Salman ; Bayu Kanigoro; Yaya Heryadi
Weather Forecasting using Deep Learning Techniques
[17] Munmun Biswas, Tanni Dhoom, Sayantanu Barua.
“Weather Forecast Prediction: An Integrated Approach
for Analyzing and Measuring Weather Data”.
[18] Mehrnoosh Torabi, Sattar Hashemi, “A Data Mining
Paradigm to Forecast Weather”, The 16th CSI
International Symposium on Artificial Intelligence and
Signal Processing (AISP 2012),IEEE, pp 579-584

IRJET- Weather Prediction for Tourism Application using ARIMA

More Related Content

What's hot (14)

Similar to IRJET- Weather Prediction for Tourism Application using ARIMA (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Weather Prediction for Tourism Application using ARIMA