Scribd
Upload
22 views

Softcom-Assignment2

Uploaded by

Yousuf ali Safin
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
22 views

Softcom-Assignment2

Uploaded by

Yousuf ali Safin
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 16

C OU R SE N O : C SE 4 1 3 8

Cou rse Tit le : Soft compu t in g

Regression Analysis for Weather Forecasting: A Comparative


Study
Research Participants

Adibul Haque Yousuf Ali Miftahul Sheikh


ID: 20200204029 ID: 20200204037 ID: 20200204038

Slide 02
Research Paper Presentation
Outline
01 Abstract 07 Evaluation Metrics

02 Introduction 08 Research Gap

03 Literature Review 9 Conclusion

04 Datasets 10 Contribution of group members

05 Pre-processing Techniques 11 References

06 Methodology

Slide 03
ABSTRACT

• Focus on rainfall prediction using ML and DL


models.

• Compared ML and DL models: SVR, LR, MLR,


LSTM, and CNN.

• LSTM and CNN achieved the highest accuracy


and lowest errors.

• LSTM showed the best overall accuracy.

Slide 04
INTRODUCTION

• Rainfall prediction is vital for agriculture,


disaster management, and urban
planning.

• Comparing regression models is key to


improving forecasting accuracy and
decision-making.

• ML and DL enhance predictions by


capturing complex weather patterns.

Slide 05
LITERATURE REVIEW

ANALYSIS OF WEATHER DATA USING • SVM showed the lowest RMSE in weather data analysis
VARIOUS REGRESSION (training: 0.0026; cross-validation: 0.0029).
ALGORITHMS. • LR showed the lowest RMSE of 0.377.

• Utilized Support Vector Regression (SVR) for rainfall


A SUPPORT VECTOR REGRESSION
prediction in Bangladesh, achieving accurate results.
MODEL FOR FORECASTING
• RMSE: 18.60 (1-day) and 24.95 (7-day); MAE: 0.18 (1-
RAINFALL.
day) and 4.99 (7-day).

Slide 06
LITERATURE REVIEW

• Mean absolute error (MAE) was around 1.9.


USING LINEAR REGRESSION FOR
• Best predictions achieved with 6 data points for 1-day
WEATHER PREDICTION.
and 72 for 5-day forecasts, highlighting the need for
targeted data selection.

• Multiple Linear Regression (MLR) achieved the highest R²


RAINFALL PREDICTION USING MACHINE score of 0.9950.
LEARNING • Lasso Regression (0.9910), XGBoost (0.9819), and
Random Forest (0.9737) also performed well.

Slide 07
DATASETS

Source: Vaibhav Mehta’s repository


with 3,271 data points.

Distribution:
• 18.03% rain data
• 81.97% no rain data
• Columns: 20 features, including
the target (rain or no rain).

Slide 08
PRE-PROCESSING TECHNIQUES

01 02 03

HANDLING OUTLIER F E AT U R E
M I SS I N G D ATA DETECTION & ENGINEERING
R E M OVA L

04 05 06

N O R M A L I Z AT I O N ENCODING HANDLING
& SCALING C AT E G O R I C A L CLASS
VA R I A B L E S IMBALANCE

Slide 9
METHODOLOGY

Machine Learning
Models:
Dataset Preprocessi • LR
ng • SVR
• MLR

Deep Learning
Model:
• LSTM

Result
Analysis

Slide 10
Result
Analysis

Slide 11
RESEARCH GAP

Paper [1]: Paper [3]:


1 3 • Linear regression methods are less precise than
• Lacks focus on real-time deployment for large
global atmospheric models.
datasets.
• More complex models or features could improve
• Doesn't evaluate model scalability for real-world
predictions.
use.

Paper [2]: Paper [4]:


2 • The study uses data only from the Chittagong 4 • More complex and hybrid models are needed to
area, limiting its generalizability. enhance prediction accuracy.
• Preprocessing is restricted to Moving Average • Advanced techniques like time series analysis
and Windowing. and clustering are underutilized in larger
datasets, impacting performance

Slide 12
CONCLUSION

LSTM provided the best results in terms of


1 accuracy and error rates.

2 CNN showed relatively poorer performance.

MLR and LR models also performed well with


3 lower error rates.

Slide 13
Related
Papers
Jahnavi, Yeturu. ”Analysis of weather data using various Sharapov, R. V. ”Using linear regression for weather
01 regression algorithms.” International Journal of Data prediction.” In 2022 Wave Electronics and its Application
Science 4, no. 2 (2019): 117 141.
03
in Information and Telecommuni cation Systems
(WECONF), pp. 1-4. IEEE, 2022.

Hasan, Nasimul, Nayan Chandra Nath, and Risul Islam Santosh Singh, Vishal Pandey, Abhishek Singh, “Rainfall
Rasel. ”A support vector regression model for forecasting Prediction Using Machine Learning,” Thakur College of
02 04 Science & Commerce, Thakur Village, Kandivali East,
rainfall.” In 2015 2nd international conference on
electrical information and communication technologies Mumbai 400101, Maharashtra, India. 2024 IJCRT, Volume
(EICT), pp. 554-559. IEEE, 2015. 12, Issue 3 March 2024, ISSN: 2320-2882.

Slide 14
CONTRIBUTION OF GROUP MEMBERS
Wr i ti n g Rep or t Prep arin g
Pap er P resen tation

Abstract, Introduction, Adib


Yousfu Ali
Conclusion.

Adibul Literature Review,


Mifta
Haque Datasets, References

Pre-processing
Miftahul
Techniques, Models, Yousuf
Sheikh
Evaluation Metrics

Slide 15
THANK YOU

You might also like