An Investigation of Weather Forecasting using Machine Learning Techniques

International Journal of Engineering and Management Research e-ISSN: 2250-0758 | p-ISSN: 2394-6962
Volume-11, Issue-1 (February 2021)
www.ijemr.net https://doi.org/10.31033/ijemr.11.1.11
72 This Work is under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
An Investigation of Weather Forecasting using Machine Learning
Techniques
Mrs. N.Vanitha1
and J.Haritha2
1
Assistant Professor, Department of Information Technology, Dr. N.G.P Arts and Science College, Coimbatore, INDIA
2
Student, Department of Information Technology, Dr. N.G.P Arts and Science College, Coimbatore, INDIA
Corresponding Author: vanitha969@gmail.com
ABSTRACT
Customarily, climate expectations are performed
with the assistance of enormous complex models of material
science, which use distinctive air conditions throughout a
significant stretch of time. In this paper, we studied a
climate expectation strategy that uses recorded
information from numerous climate stations to prepare
basic AI models, which can give usable figures about
certain climate conditions for the not so distant future
inside a brief timeframe These conditions are frequently
flimsy on account of annoyances of the climate framework,
making the models give mistaken estimates.[1] The model
are for the most part run on many hubs in an enormous
High Performance Computing (HPC) climate which burns
through a lot of energy.. The modes can be run on
significantly less asset serious conditions. In this paper we
describe that the sufficient to be utilized status of the
workmanship methods. Moreover, we described that it is
valuable to use the climate stations information from
various adjoining territories over the information of just
the region for which climate anticipating is being
performed.
Keyword-- Weather Forecasting, Machine Learning,
Types, Methods, Significance, Technique
I. INTRODUCTION
AI (ML) is the investigation of PC calculations
that improve naturally through experience. It is viewed
as a piece of man-made consciousness. [2]AI
calculations construct a model dependent on example
information, known as "preparing information", to settle
on expectations or choices without being expressly
customized to do so. Machine learning calculations are
utilized in a wide assortment of utilizations, for example,
email sifting and PC vision, where it is troublesome or
impractical to create customary calculations to play out
the required undertaking.
1.1 Types of Machine Learning
Machine Learning is mainly classified into
three types, they are mentioned in figure1.
 Supervised Learning
 Unsupervised Learning
 Reinforcement Learning
Figure 1: Classification of Machine Learning
Supervised Learning - Portrays a class of issue that
includes utilizing a model to get familiar with a planning
between input models and the objective variable.
Unsupervised Learning - Portrays a class of issues that
includes utilizing a model to depict or remove
connections in information.
Reinforcement Learning - Depicts a class of issues
where a specialist works in a climate and should figure
out how to work utilizing criticism.
1.2 Significance of Machine Learning
Here some of the importance of the machine
learning is mentioned below[3].
MACHINE LEARNING
SUPERVISED
UNSUPERVISED
REINFORCEMENT

 Blogger's information is classified into expert
and occasional bloggers utilizing AI
calculations.
 Choice tree calculations, apathetic learning
strategies and resembling procedures are
applied to standard datasets to investigate and
analyse the information characterization
calculations on ostensible information.
 Arbitrary Forest and Nearest-neighbour
classifiers accomplished high precision for
information grouping.
 Significance of choice tree rules for factor
distinguishing proof behind the polished
methodology of bloggers is explained.
1.3 Application of Machine Learning
Some of the popular machines learning
application are mentioned below in figure 2.
Figure 2: Application of Machine Learning
II. INTRODUCTION ON WEATHER
FORECASTING
The way toward foreseeing climate conditions
for future is called climate determining.[4] This paper
gives the downpour expectation the utilization of
ongoing information of temperature, dampness, and
pressing factor utilizing different sensors. When
determined by hand dependent on upon changes in
barometric pressing factor, current climate conditions,
and sky condition or overcast cover, climate gauging
now depends on PC based models that consider
numerous air factors. Climate figures are made by
gathering however much information as could
reasonably be expected about the present status of the air
(especially the temperature, dampness and wind) and
utilizing comprehension of air measures (through
meteorology) to decide how the air advances later on.
Notwithstanding, the turbulent idea of the climate and
fragmented comprehension of the cycles imply that
conjectures become less precise as the scope of the
estimate increments.
2.1Methods of Weather Forecasting
Obviously for climate gauge to exist there
should be techniques on which it is finished. These
techniques are as per the following:
 Persistence Forecasting
 Climatology Forecasting
 Looking at the sky
 Use of a barometer
 Now casting
 Utilization of Forecasting Models
 Analogue Forecasting
Persistence Forecasting
Constancy anticipating is the most
straightforward technique for gauging which expects a
continuation of the present. It depends upon the present
conditions to gauge the climate when it is consistent
state, for example, throughout the late spring season in
the jungles.[5] This strategy for anticipating
unequivocally relies on the presence of a stale climate
design. It very well may be valuable in both short-range
gauges and long-range conjectures.
Climatology Forecasting
Climatology estimate depends on the perception
that climate for a specific day at an area doesn't change
much starting with one year then onto the next. Thus, a
drawn out normal of climate on a specific day or month
ought to be a decent speculation as the climate for that
day or month. The clearest climatology figure in this
piece of the world (Nigeria) is, "Cold in December,
warm in July (the well-known July break)". One
shouldn't be a meteorologist to make that estimate.
Looking at the Sky
Alongside pressure propensity, utilization of the
sky condition is one of more significant climate
boundaries that can be utilized to conjecture climate in
rugged territories. Thickening of overcast cover or the
attack of a higher cloud deck is demonstrative of
downpour soon. Morning mist forecasts reasonable
conditions, as stormy conditions are gone before by wind
or mists, which forestall haze development. The
methodology of a line of rainstorm could demonstrate
APPLICATION OF MACHINE
LEARNING
Medical
Prediction
Banking and stock Speech recognition
Classification Image recognition
Regression

the methodology of a virus front. Cloud free skies are
demonstrative of reasonable climate for the not distant
future. [6]The utilization of sky cover in climate
expectation has prompted different climate legend
throughout the long term.
Use of a Barometer
Utilizing barometric pressing factor and the
pressing factor inclination (x the difference in pressing
factor after some time) has been utilized in estimating
since the late nineteenth century. The bigger the
adjustment in pressing factor, particularly, if more than
2.54mmHg, the bigger the adjustment in climate can be
normal. On the off chance that the pressing factor drop is
fast, a low-pressure framework is drawing nearer, and
there is a more prominent possibility of downpour.
Quick pressing factor rises are related with improving
climate conditions, for example, clearing skies.
Nowcasting
The anticipating of the climate inside the
following six hours is regularly alluded to as nowcasting.
In this time range, it is conceivable to figure more
modest highlights, for example, singular showers and
tempests with sensible exactness, just as different
highlights too little to be in any way settled by a PC
model. A human given the most recent radar, satellite
and observational information will have the option to
improve an examination of the limited scale highlights
present thus will have the option to make a more exact
estimate for the accompanying not many hours.
Utilization of Forecasting Models
Previously, the human forecasters were
responsible for delivering the entire environment guess
reliant on open discernment. Today, human data is
overall confined to picking a model reliant on various
limits, for instance, model tendencies and execution.
Using an understanding of guess models, similarly as
outfit people from the various models, can help decline
check botch. Regardless, despite how minimal the
typical bungle becomes with any individual system, huge
errors inside a particular piece of course are up 'til now
possible on some arbitrary model run.
Analogue Forecasting
The simple strategy is a perplexing method of
making a conjecture, requiring the forecaster to recollect
a past climate occasion which is relied upon to be copied
by an impending occasion. The simple forecaster's
assignment is to find the date in history when the climate
is an ideal match, or simple, to the present climate. At
that point the estimate for later is straightforward –
whatever occurred in the day after the simple will be the
climate for later.
Ensemble Forecasting
Albeit an estimate model will foresee climate
highlights developing practically into the removed
future, the mistakes in a conjecture will definitely
develop with time because of the clamorous idea of the
air and the inaccuracy of the underlying perceptions. The
detail that can be given in a conjecture in this manner
diminishes with time as these mistakes increment. These
become a moment that the mistakes are enormous to the
point that the figure has no relationship with the genuine
condition of the environment.
2.2 Application of Weather Forecasting
There are many types of application were
available in weather forecasting. [7]Some of them are
mentioned below using diagrammatic representation.
 Agriculture
 Utility companies
 Private sector and military applications
 Marine
 Severe weather alerts and advisories
 Air Traffic
Figure 3: Application of Weather Forecasting
APPLICATION OF WEATHER
FORECASTING
Agriculture
Utility companies
Private sector
and military
applications
Marine Air Traffic
Severe weather alerts
and advisories

2.3 Significance of Weather Forecasting
Importance of weather forecasting is clearly
mentioned below[8].
 Estimating and anticipating the climate has the
ability to help individuals, organizations,
ranchers, transport frameworks and give
cautioning frameworks. It is likewise significant
in deciding a region's environment, which
includes estimating the climate throughout an
extensive stretch of time.
 These stations empower of climate conditions,
for example, wind speed and course, stickiness,
precipitation, temperature, and barometric
pressing factor.
2.4 Significance of Machine Learning in Weather
Forecasting
Some of the importance of machine learning in
weather forecasting is clearly discussed below.
 Observing these conditions makes it
conceivable to anticipate climate and give
significant information to the National Weather
Service and nearby meteorologists.
 The usage of AI in forecast of climate
conditions in brief timeframes, which can run
on less asset escalated machines.
 Usage of mechanized frameworks to gather
recorded information from a committed climate
administration. Thorough evaluation of the
proposed technique and comparison of several
machine learning models in the prediction of
future weather conditions.
III. BACKGROUND STUDY
In 2019, Nitin Singh, Saurabh Chaturvedi,
Shamim Akhter published a journal in IEEE Xplore on
the topic of Weather Forecasting Using Machine
Learning Algorithm using the method of Raspberry pie 3
B on Python and given the outcome of the prime
objective of this work is to develop a low cost, reliable,
and efficient weather forecasting application using the
ML concept in python on Raspberry pi board.
In2018,A H M Jakaria, Md Mosharaf Hossain,
Mohammad Ashiqur Rahman published a journal in
IEEE Xplore on the topic of Smart Weather Forecasting
Using Machine Learning using the method of Random
Forest Regression, SVR and ETR and given the outcome
of in that section, they present a thorough evaluation of
their models trained with the weather station data.
In 2018, Sue Ellen Haupt, Jim Cowie, Seth
Linden, Tyler McCandless, Branko Kosovic, Stefano
Alessandrini published a journal in IEEE Xplore on the
topic of Machine Learning for Applied Weather
Prediction using the method of NCAR’s Sun4Cast Solar
power prediction system and given the outcome of Using
that method is just one of many emerging applications of
applied weather forecasting that blends the best of their
knowledge of physics, numerics, and AI using smart big
data and leveraging the internet of things.
In 2020, K. Geetha Rani, Dr D. C. Joy Winnie
Wise, S. Sufiyah Begum, S. Nirosha published a journal
in IEEE Xplore on the topic of Designing A Model for
Weather Forecasting Using Machine Learning using the
method of POTEKA sensor to watched information
consistency through portable system and given the
outcome of the weather statistics module displays a
visual presentation of the weather for a particular time
series upon the entering of the input data. Based on this,
the user can calculate the cumulative weather of the
particular region over a given time period.
In 2019, Anjali T, Chandini K, Anoop K, Lajish
V L published a journal in IEEE Xplore on the topic of
Temperature Prediction Using Machine Learning
Approach using the method of Temperature prediction
using Multiple Linear Regression, Artificial Neural
Network, Support Vector Machine and given the
outcome of the observation concludes that among all the
approaches MLR and regression based SVM gives better
results. The results also imply the need for a hybrid
approach to improve the accuracy of this proposed
approach.
In 2011, Navin Sharma, Pranshu Sharma, David
Irwin, and Prashant Shenoy published a journal in IEEE
Xplore on the topic of Prediction Solar Generation from
Weather Forecasts Using Machine Learning using the
method of Least Linear Square Regression, Support
Vector Machine, Elimination Redundant Information,
Comparing with Existing Models and given the outcome
of the results indicate that automatically generating
accurate models that predict solar intensity, and hence
energy harvesting of solar arrays, from weather forecasts
is a promising area and data centres that utilize on-site
solar arrays to generate power.
Table 1: Background Study
SI NO AUTHOR JOURNAL TITLE OF THE PAPER METHOD AND
YEAR
OUTCOME
1 Nitin Singh, Saurabh
Chaturvedi, Shamim Akhter
IEEE
XPLORE
Weather Forecasting Using
Machine Learning Algorithm
Using Raspberry pie 3
B on Python in 2019.
The prime objective of this
work is to develop a low cost,
reliable, and efficient weather
forecasting application using
the ML concept in python on
Raspberry pi board.
2 A H M Jakaria, Md
Mosharaf Hossain,
IEEE
XPLORE
Smart Weather Forecasting
Using Machine Learning
Random Forest
Regression, SVR and
In that section, they present a
thorough evaluation of their

Mohammad Ashiqur
Rahman
ETR in 2018. models trained with the
weather station data.
3 Sue Ellen Haupt, Jim
Cowie, Seth Linden, Tyler
McCandless, Branko
Kosovic, Stefano
Alessandrini
IEEE
XPLORE
Machine Learning for Applied
Weather Prediction
NCAR’s Sun4Cast
Solar power prediction
system in 2018.
Using this method is just one
of many emerging
applications of applied
weather forecasting that
blends the best of their
knowledge of physics,
numerics, and AI using smart
big data and leveraging the
internet of things.
4 K. Geetha Rani, Dr D. C.
Joy Winnie Wise, S.
Sufiyah Begum, S.Nirosha
IEEE
XPLORE
Designing A Model for Weather
Forecasting Using Machine
Learning
Using POTEKA sensor
to watched information
consistency through
portable system in
2020.
The weather statistics module
displays a visual presentation
of the weather for a particular
time series upon the entering
of the input data. Based on
this, the user can calculate the
cumulative weather of the
particular region over a given
time period.
5 Anjali T, Chandini K,
Anoop K, Lajish V L
IEEE
XPLORE
Temperature Prediction Using
Machine Learning Approaches
Temperature prediction
using Multiple Linear
Regression, Artificial
Neural Network,
Support Vector
Machine in 2019.
The observation concludes
that among all the approaches
MLR and regression based
SVM gives better results. The
results also imply the need for
a hybrid approach to improve
the accuracy of this proposed
approach.
6 Navin Sharma, Pranshu
Sharma, David Irwin, and
Prashant Shenoy
IEEE
XPLORE
Prediction Solar Generation
from Weather Forecasts Using
Machine Learning
Least Linear Square
Regression, Support
Vector Machine,
Elimination Redundant
Information,
Comparing with
Existing Models in
2011.
The results indicate that
automatically generating
accurate models that predict
solar intensity, and hence
energy harvesting of solar
arrays, from weather forecasts
is a promising area and data
centres that utilize on-site
solar arrays to generate
power.
This table indicates that the clear analysisof literature on machine learning for weather forecasting.
Table2: Analysis of literature on machine learning for weather forecasting
Authors/Methods Raspberry pi
Random
Forest
Regression,
SVR
NCAR’s
Sun4Cast
Solar power
prediction
system
POTEKA
sensor
Multiple
Linear
Regression,
Artificial
Neural
Network
Support
Vector
Machine
1.Nitin Singh,
Saurabh
Chaturvedi,
Shamim Akhter
 - - - - -
2.A H M Jakaria,
Md Mosharaf
Hossain,
Mohammad
Ashiqur Rahman
-  - - - -
3.Sue Ellen Haupt,
Jim Cowie, Seth
Linden, Tyler
McCandless,
Branko Kosovic,
Stefano
- -  - - -

Alessandrini
4.K. Geetha Rani,
Dr D. C. Joy
Winnie Wise, S.
Sufiyah Begum,
S.Nirosha
- - -  - -
5.Anjali T,
Chandini K, Anoop
K, Lajish V L
- - - -  
6.Navin Sharma,
Pranshu Sharma,
David Irwin, and
Prashant Shenoy
- - - -  
IV. RESEARCH GAP IDENTIFIED
 Most of the research studied, In this literature is
focussed on the local analysis of the weather
prediction. However there is not an extensive
study about the weather prediction of
temperature at a global level by means of these
ML based approaches. Taking into an account
the robust data currently available in devised
website, different ML strategies an input
features could be used to accurately predict
temperature at global level.
 Research reported at the regional level has no
deeply analyzed the dependency of the
temperature values of the surroundings area in
the temperature estimation. A study oriented to
analyze the impact of using temperature values
of surrounding a station as inputs, based on the
distance each other, could be of particular
interest.
 A large number of the words described in this
paper do not include time horizon analysis, The
lack of these information makes it difficult to
have a better idea of the accuracy of the method
proposed.
V. CONCLUSION AND FUTURE
WORK
In this work, we studied many machine learning
algorithms to detect weather forecasting techniques in
future we will try on all features of machine learning
techniques and to achieve best accuracy. Our
investigation likewise energizes that nobody procedure
can be mentioned being the ideal AI strategy. Hence
there is a solid requirement for better understanding into
the legitimacy and consensus of many talked about
methods.
REFERENCES
[1] Abraham, Ajith, Ninan Sajith Philip, Baikunth Nath,
& P. Saratchandran. (2002). Performance analysis of
connectionist paradigms for modeling chaotic behavior
of stock indices." In: Second International Workshop on
Intelligent Systems Design and Applications,
Computational Intelligence and Applications, Dynamic
Publishers Inc., USA, pp. 181-186.
[2] Goddard, Lisa, Simon J. Mason, Stephen E. Zebiak,
Chester F. Ropelewski, Reid Basher, & Mark A. Cane.
(2001). Current approaches to seasonal to interannual
climate predictions. International Journal of
Climatology, 21(9), 1111-1152.
[3] Chaves, Rosane R., Robert S. Ross, & T. N.
Krishnamurti. (2005). Weather and seasonal climate
prediction for South America using a multi model super
ensemble. International Journal of Climatology, 25(14),
1881-1914.
[4] Baboo, S. Santhosh, I., & Kadar Shereef. (2010). An
efficient weather forecasting system using artificial
neural network. International Journal of Environmental
Science and Development, 1(4), 321.
[5] Paras, Sanjay Mathur. (2016). A simple weather
forecasting model using mathematical regression. Indian
Research Journal of Extension Education, 12(2), 161-
168.
[6] N. Hasan, M. T. Uddin, & N. K. Chowdhury. (2016).
Automated weather event analysis with machine
learning. In: Proc. IEEE2016 International Conference
on Innovations in Science, Engineering and Technology
(ICISET), pp. 1-5.
[7] L. L. Lai, H. Braun, Q. P. Zhang, Q. Wu, Y. N. Ma,
W. C. Sun, & L. Yang. (2004). Intelligent weather
forecast. In: Proc. IEEE 2004 International Conference
on Machine Learning and Cybernetics, pp. 4216-4221.
[8] A. G. Salman, B. Kanigoro, & Y. Heryadi. (2015).
Weather forecasting using deep learning techniques. In:
Proc.IEEE2015 International Conference on Advanced
Computer Science and Information Systems (ICACSIS),
pp. 281-285.
[9] Delhi Weather Data. [Online].
Available at:
https://www.kaggle.com/mahirkukreja/delhi-
weatherdata/home.
[10] Raspberry Pi. [Online]. Available
at: https://en.wikipedia.org/wiki/Raspberry_Pi.
[11] Database of State Incentives for Renewables and
Efficiency. (2010). Available at:
http://www.dsireusa.org.

[12] State of California Executive Order S-21-09.
(2009). Available at: http://gov.ca.gov/executive-
order/13269.
[13] Freeing the grid: Best and worst practices in state
net metering policies and interconnection procedures.
(2009). Available at:
http://www.newenergychoices.org/uploads/FreeingTheG
rid2009.pdf.
[14] N. Sharma, J. Gummeson, D. Irwin, & P. Shenoy.
(2010 June). Cloudy computing: leveraging weather
forecasts in energy harvesting sensor systems. In:
SECON.
[15] N. Cristianini & J. Shawe-Taylor. (2000). An
introduction to support vector machines and other
kernel-based learning methods. Cambridge University
Press.
[16] https://www.wunderground.com/weather/api.
[17] Weather forecast using the sensor data from your
IoT hub in Azure Machine Learning. (2018). Available
at: https://docs.microsoft.com/en-us/azure/iot-hub/ iot-
hub-weather-forecast-machine-learning.
[18] Dan Becker. (2018). Using categorical data with
one hot encoding. Available at:
https://www.kaggle.com/dansbecker/using-categorical-
data-with-one-hot-encoding/.
[19] Jeffrey Burt. (2017). Machine learning storms into
climate research. Available at:
https://www.nextplatform.com/2017/04/18/machine-
learning-storms-climate-research/.
[20] Aditya Grover, Ashish Kapoor, & Eric Horvitz.
(2015). A deep hybrid model for weather forecasting. In:
Proceedings of the 21th ACM SIGKDD International
Conference on Knowledge Discovery and Data Mining.
ACM, pp. 379–386.

An Investigation of Weather Forecasting using Machine Learning Techniques

More Related Content

What's hot (20)

Similar to An Investigation of Weather Forecasting using Machine Learning Techniques (20)

More from Dr. Amarjeet Singh (20)

Recently uploaded (20)

An Investigation of Weather Forecasting using Machine Learning Techniques