Scribd
Upload
10 views2 pages

Lecture 298

Uploaded by

Sparsh Soni
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
10 views2 pages

Lecture 298

Uploaded by

Sparsh Soni
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

302 M. T. Nuseir et al.

Table 1 Comparison of previous model results


S. No. Authors Model RMSE (%)
1 Yen et al. (2019) Deep ESN Model (Echo state 1.51
Network)
2 Khashei et al. (2021) Multiple Linear Regression Model 26.5
3 Shah et al. (2018) Artificial Neural Network based 68.49
Model
4 Reddy et al. (2022) Integrated Model (Singular Spectrum 0.716
Analysis (SSA)—Least-Squares
Support Vector Regression (LSSVR),
SSA—Least Square Random Forest
(LSRF))
5 Diez-Sierra and del Jesus (2020) Support Vector Machine (SVM) 15.2

5 Empirical Analysis

Recently, rainfall forecast is the main issue for climatological division as it is directly
linked to the financial system and life of people. It is a reason for ecological disasters
like floods and famine which are faced by people across the world every year. The
precision of rainfall forecasting has good importance for such countries as Pakistan,
India, Australia, and Canada whose financial system is mostly reliant on agriculture.
In this research, analysis is done on some previous models which calculates Root
Mean Square Error (RMSE) in rainfall prediction, as shown in Table 1. Because of
the dynamic nature of the environment, Statistical techniques fail to deliver good
accuracy for rainfall forecasting. Recent statistical techniques must perform well in
predict of rainfall in terms of accuracy. Therefore, this research proposed machine
learning-based statistical tools that predict rainfall accurately with low RMSE.

6 Discussion

As per the above discussion, in yearly rainfall data, there is no simple method for
determination of the rainfall parameters such as wind speed, humidity, soil temper-
ature, etc. Too few or too many input parameters can affect either the learning or
prediction capability of the network. Users cannot use the same model over a long
period because parameters are varying from day to day, month to month, or year to
year. Therefore, new parameters cannot be fitted into the developed model. Different
types of NN modeling for yearly, monthly, and weekly rainfall data are described as
follows. In yearly rainfall prediction, average rainfall data is used as input to ANN.
Therefore, ANN can easily predict the approximate peak value of yearly rainfall
data.
Machine Learning Based Statistical Tools Estimation for Rainfall … 303

7 Conclusion

This paper discussed various ML approaches for rainfall forecasting, and issues
requiring attention while applying these approaches for rainfall prediction and
other forecasting techniques such as statistical and numerical methods to get better
results. In this study, several machine learning (ML) methods, including Multiple
Linear Regression, Integrated technique (Singular Spectrum Analysis (SSA)—Least-
Squares Support Vector Regression (LSSVR), SSA—Least Square Random Forest
(LSRF)), Artificial Neural Networks (ANN), Deep ESN technique (Echo state
Network), Random Forest (RF), and Support Vector Machines (SVM) are discussed
with RMSE but they don’t show better performance in rainfall forecasting. The
proposed model of this study may show state of the art performance for rainfall
forecasting which is very essential for such countries whose economy is depend on
agriculture sector.

8 Limitations and Future Directions

Macroclimatic and climatic influences, including such continentally and height


impacts, as well as ordinary precipitation volume, have a serious influence on the
ANNs’ predicting accuracy more well seasons. The biggest constraints are extended
forecasting because algorithms split as they start expanding into the future. The
above-proposed model predicts better results with the help of ML algorithms but is
not uncommon for one model to have a high-pressure overhead with dry weather
while another forecast has a low pressure with precipitation. The time horizon has
been a lot more realistic than in the prior, with our short-term forecasts exploding in
accuracy. The future advice remains the need for greater data and computing power
to make better decisions.

References

Aftab, S., Ahmad, M., Hameed, N., Bashir, M. S., Ali, I., & Nawaz, Z. (2018a). Rainfall prediction
using data mining techniques: A systematic literature review. International Journal of Advanced
Computer Science and Applications, 9, 143–150.
Aftab, S., Ahmad, M., Hameed, N., Bashir, M. S., Ali, I., & Nawaz, Z. (2018b). Rainfall prediction
in Lahore City using data mining techniques. International Journal of Advanced Computer
Science and Applications, 9, 254–260.
Ahmad, M., & Aftab, S. (2017). Analyzing the performance of SVM for polarity detection with
different datasets. International Journal of Modern Education and Computer Science, 9, 29–36.
Al-Tahat, S., & Moneim, O. A. (2020). The impact of artificial intelligence on the correct applica-
tion of cyber governance in Jordanian commercial banks. International Journal of Scientific &
Technology Research, 9, 7138–7144.

You might also like