PREDICTIVE MODELLING OF ELECTRICITY PRICES USING

MACHINE LEARNING TECHNIQUES

Report submitted to the SASTRA Deemed to be University

as the requirements for the course

B.Tech., Computer Science and Engineering

CSE300 - MINI PROJECT

Submitted by

ARCHANA K (225003015)
BHUVANESWARI S (225003022)
SHAKTHI K (225003133)

MAY 2024

Department of Computer Science and Engineering

SRINIVASA RAMANUJAN CENTRE
Kumbakonam, Tamil Nadu, INDIA -612001

MAY 2024

i
 ACKNOWLEDGEMENT

I would like to thank our Honorable Chancellor Prof. R. Sethuraman for providing us
with an opportunity and the necessary infrastructure for carrying out this project as a
part of our curriculum.
I would like to thank our Honorable Vice-Chancellor Dr. S. Vaidhyasubramaniam
and Dr. S. Swaminathan, Dean, Planning & Development, for the encouragement and
strategic support at every step of our college life.

I extend our sincere thanks to Dr. R. Chandramouli, Registrar, SASTRA Deemed to be

University for providing the opportunity to pursue this project.
I extend our heartfelt thanks to Dr. V. Ramaswamy, The Dean and Dr. A. Alli Rani,
Associate Dean, Srinivasa Ramanujan Centre for their constant support and
suggestions when required without any reservations.

I express my sincere gratitude to Dr. V. Kalaichelvi, Associate Professor, Department

of Computer Science and Engineering for her unconditional support and
encouragement for completing the project.
My guide Prof.A.Menaga, Assistant Professor, Department of Computer Science &
Engineering, Srinivasa Ramanujan Centre was the driving force behind this whole idea
from the start. Her deep insight in the field and invaluable suggestions helped me in
making progress throughout our project work.

I also thank the Project Coordinator Dr. S. Priyanga, Assistant Professor, Department
of Computer Science & Engineering, Srinivasa Ramanujan Centre for her ever-
encouraging spirit and meticulous guidance for the completion of the project. We also
thank the panel members for their valuable comments and insights which made this
project better.
I would like to extend our gratitude to all the teaching and non-teaching faculties of the
Srinivasa Ramanujan Centre who have either directly or indirectly helped us in the
completion of the project. I gratefully acknowledge all the contributions and
encouragement from my family and friends resulting in the successful completion of
this project. I thank you all for providing me with an opportunity to showcase my skills
through the project.

iii
 LIST OF FIGURES

Figure No Title Page No

1.1 Workflow 4
1.2 Data Preprocessing 7
1.3 Principal component 11
analysis
1.4 Probilistic model 12
1.5 Ensemble model 14
1.6 Visualization loss 15
Curve performance
1.7 t-SNE 16
1.8 ICA 19
1.9 Dimensionality reduction 29
screenshots
1.10 Logistic resgression 23

1.11 Result of models

30
1.12 Output screenshot 31
1.13 Exporting File at Desired 33
Location

iv
 ABBREVIATIONS

MSE Mean Square Error

R2 SCORE R Squared Score

PCA Principal Component Analysis

t-SNE t-distributed Stochastic Neighbor Embedding

ICA Independent Component Analysis

v
 ABSTRACT

Modern energy markets are complex systems influenced by a wide range of factors, including supply
and demand dynamics, weather conditions, and geopolitical events. As a result, electricity prices can
exhibit high volatility, making accurate forecasting a challenging task. For energy producers, grid
operators, and consumers, however, accurate price predictions can lead to significant cost savings
and more efficient utilization of resources.
In this project, our goal is to develop a machine learning model that accurately predicts electricity
prices based on a dataset containing information on energy consumption, generation, and other
relevant factors. Our target variable for prediction will be the actual price of electricity. In this
project, we are going to implement machine learning models.

KEYWORDS: Energy Bills, Energy Savings, Predictive modeling, Electricity consumption,

Carbon emission, Predict electricity prices, Machine learning.

Specific Contribution:

ARCHANA K - PPT creation, probabilistic model, Dataset

analysis, and findings.
BHUVANESWARI S- Data preprocessing, Ensemble model
SHAKTHI K - Dimensionality reduction, Logistic regression
model, Deployment

Specific Learning:

ARCHANA K - Probabilistic model

BHUVANESWARI S - Ensemble model
SHAKTHI K - Logistic Regression
Technical Limitations & Ethical Challenges Faced:

➢ For real-time applications data should have high-resolution

➢ Data availability, quality, and quantitative measure

vi
 TABLE OF CONTENTS

Title Page No:

Bonafide Certificate ii
Acknowledgment iii
List of Figures iv
Abbreviations v
Abstract vi
1. Summary of the base paper 1
1.1 Introduction
1.2 Proposed System
1.3 Model Implementation
1.4 Dataset
2. Merits and Demerits of the base paper 5
2.1 Literature Review
2.2 Merits
2.3 Limitations

3. Source Code 7
4. Output Snapshots 31
5. Conclusion and Future Work 35
6. References 36
7. Appendix - Base Paper 46

vii
 CHAPTER - 1

SUMMARY OF THE BASE PAPER

Title:
Winter demand falls as fuel bills rise: Understanding the energy impacts of the cost-of-
living crisis in British households

Journal Name: Energy & Buildings

Publisher Name:
Ellen Zapata-Webborn, Clare Hanmer, Tadj Oreszczy, Gesche Huebner,
Eoghan McKenna, Jessica Few, Simon Elam, Martin Pullinger, Callum Cheshire,
Dominic Friel,Harry Masters, Alex Whittaker
Year: 2024

1.1 Introduction:
Predicting electricity prices is challenging due to the inherent volatility and non-
linearity of energy markets. Several factors influence electricity prices, including
demand patterns, weather conditions, fuel prices, regulatory policies, market dynamics,
and renewable energy penetration. Moreover, electricity markets exhibit complex
temporal and spatial dependencies, making traditional forecasting methods less
effective.
Electricity price forecasting is a crucial task for various stakeholders in the
energy sector, including utility companies, power traders, policymakers, and
consumers. Accurate predictions of electricity prices enable efficient decisionmaking,
resource planning, and risk management. Machine learning techniques have emerged
as powerful tools for electricity price prediction

1.2 Proposed System:

• Predict the electricity prices by including features from the dataset, considering the year
as input to predict the prices for that specific year.

• Prediction is done by using machine learning technique algorithms

1
Data Collection:
It involves gathering relevant datasets that capture the factors that influence
electricity prices. Some of the factors are generation biomass, wind onshore, wind
offshore, solar, renewable, etc,

Data pre-processing:
It is done by various techniques to clean, transform, and format the data
to make it suitable for machine learning algorithms. Identify and handle missing values
in the dataset. it can be solved by imputing mean values to the missing data.

Data augmentation:

This is done to improve the performance and robustness of models in predicting

electricity prices. By generating synthetic data from original datasets the model can
learn more effectively and generalize better-unseen data.

1.3 Implementation process:

1. Survey Data Analysis: The paper incorporates survey responses from

households to complement the smart meter data analysis. This allows for a more
comprehensive examination of energy-saving behaviors, self-reported actions, and
perceptions related to energy consumption during the cost-of-living crisis.

2. Weather Normalization: By considering weather data in the analysis, the study

accounts for the impact of external factors on energy consumption variations. Weather
normalization helps isolate the effects of the cost-of-living crisis on energy demand
from seasonal fluctuations.

3. Household-Level Analysis: The research focuses on individual households,

enabling a detailed examination of energy consumption changes, energy-saving
actions, and financial struggles at a granular level. This approach provides insights into
the diverse responses and behaviors of households during the crisis.

4. Longitudinal Analysis: The study utilizes historical data from previous winters
to establish baseline energy consumption patterns and trends. By comparing these
patterns with the winter of 2022/23, the research can identify deviations and assess the
specific impacts of the cost-of-living crisis on energy usage.

5. Policy Implications: The findings of the study may have implications for energy
policy development, social welfare programs, and interventions aimed at addressing
fuel poverty and promoting energy efficiency. Understanding how households respond
to economic challenges can inform targeted policy measures.

2
6. Transparency and Reproducibility: The paper emphasizes transparency by
providing detailed information on data sources, methodology, model training
processes, and performance metrics. This commitment to transparency enhances the
reproducibility of the study and allows for validation by other researchers.

7. Interdisciplinary Approach: The study likely involves interdisciplinary

collaboration, combining expertise in energy economics, data science, behavioral
psychology, and policy analysis. This interdisciplinary approach enriches the research
by considering various perspectives and methodologies.

8. Further analysis of the results, potentially exploring the relationship between

financial well-being, energy behaviors, and the challenges faced by households in
managing energy costs during the crisis.

9. Discussion on specific energy-saving actions adopted by households, the

differences in behavior between different quintiles of households, and the potential for
reducing energy consumption through targeted interventions.

10. Detailed analysis of specific energy-saving actions, their impact on energy

consumption, and the risk ratios associated with different behaviors adopted by
households during the cost-of-living crisis.

11. Continued discussion on energy-saving actions, potentially highlighting the

most effective strategies for reducing energy consumption and addressing fuel poverty
in households.

12. Analysis of the limitations of the study, including potential biases in the data,
challenges in data interpretation, and the assumptions made in the analysis process.

13. Further discussion on the limitations of the study, potentially addressing issues
related to data availability, model assumptions, and the generalizability of the findings.

14. Conclusion summarizing the key findings of the study, implications for energy
policy and household energy management, and suggestions for future research
directions.

3
Workflow:

Figure 1.2

1.4 Dataset:

• Energy datasets can cover a wide range of topics, including electricity

consumption, renewable energy production, energy prices, and more. Our
dataset includes 28 attributes
• Our dataset has 35065 rows and 28 columns(35065x28)
• The dataset contains information on energy consumption spanning four years,
with energy values recorded for each hour throughout the duration.

4
 CHAPTER - 2

MERITS AND DEMERITS OF THE BASE PAPER

2.1 Literature Review:

Following are some of the research papers which came up with various methods for
electrcity price prediction

1.Author & year: Yishuai Lina, Zhiming Xionga, Jiawei Zhu & 2023
Title :Optimization of the electricity consumption behaviors of users under
uncertain electricity prices and consumption patterns

2. Author & year :Milena Ðukanovic, Ljiljana Kašcelan, Suncica Vukovic (Rogic),
Ivan Martinovic,Martin Calasan & 2023
Title: A machine Learning approach for time series forecasting with application to
dept risk of the Montenegrin electricity industry

3. Author & year:Martin Janos Mayer, Bence Biro, Botond Szücs, Attila Aszodi &
2023
Title :Probabilistic modeling of future electricity systems with high renewable
energy penetration using machine learning

4. Author & year:Suleman Sarwar, Ghazala Aziz, Aviral Kumar Tiwari &2023
Title :Implication of machine learning techniques to forecast the electricity price
and carbon emission: Evidence from a hot region.

5. Author & year:Younes Ledmaoui, Adila El Maghraoui, Mohamed El Aroussi,

Rachid Saadane, Ahmed Chebak, Abdellah Chehri&2023
Title: Forecasting solar energy production: A comparative study of machine
learning algorithms.

Merits and Limitations of the proposed system

2.2 Merits:

1.The use of XGBoost and predictive modeling techniques in the base paper allows
for more accurate predictions of energy consumption changes in households during the
cost-of-living crisis compared to traditional regression models or simple time series
analysis.

5
 2.Machine learning algorithms like XGBoost can handle complex relationships and
interactions between variables in the data, providing a more nuanced understanding of
the factors influencing energy consumption behaviors.

3.The proposed techniques are scalable and can be applied to large datasets, making
them suitable for analyzing energy consumption patterns across a wide range of
households.
4. While machine learning models can be complex, efforts to explain and interpret
the results, as demonstrated in the base paper, enhance the transparency and
understanding of the analysis.

2.3 Limitations:

1. Machine learning algorithms like XG Boost may require a significant amount of

data for training and validation, which could be a limitation for smaller datasets or
regions with limited smart meter infrastructure.

2. The complexity of machine learning models can make them harder to interpret and
explain, potentially leading to challenges in understanding the underlying factors
driving energy consumption changes.

3. Training and optimizing machine learning models like XG Boost can be

computationally intensive and may require specialized expertise, software, and
hardware resources.

4. There is a risk of overfitting with complex machine learning models, where the
model performs well on the training data but fails to generalize to new, unseen data.

6
3. SOURCE CODE:

DATA PREPROCESSING

7
8
9
10
11
 Figure1.3

Dimensionality Reduction :
1) PCA (Principal component analysis)

Figure1.4

12
13
Figure 1.5

14
Figure1.6

15
 Figure1.7

Figure 1.8

16
17
18
19
Figure 1.9

20
21
22
Logistic regression(Deployment model):

23
24
25
26
27
28
29
30
4. OUTPUT GRAPHS:

DIMENSIONALITY REDUCTION
Figure 1.9

31
 –

SCREENSHOTS:
32
33
 Figure1.13

34
Exporting File at Desired Location.
Figure 1.14

35
 CHAPTER - 5

CONCLUSION

Conclusion:

• Our study used various machine learning techniques to predict electricity prices. The
probabilistic model helped us not only forecast prices but also estimate uncertainties. The
ensemble model, combining multiple methods, improved accuracy by capturing complex
patterns. Unsupervised models revealed hidden insights in the data.
• The results of this study can help people who make decisions in the energy industry to plan
more effectively.
• RESULTS: https://shakthikann.github.io/energy_predict/

Future Plans:

• In this project, our goal is to develop a machine learning model that accurately predicts electricity
prices based on a dataset containing information on energy consumption, generation, and other
relevant factors. Our target variable for prediction will be the actual price of electricity.

36
 CHAPTER - 6

REFERENCES

1. Shobhit Agarwal, “What are some problems faced by deaf and dumb people while using todays
common tech like phones and PCs”, 2017 [Online].
Available:https://www.quora.com/What-are-some-problems-faced-by-deaf-and-dumb-
people-whileusing-todays-common-tech-like-phones-and-PCs,[Accessed April 06, 2019].

2. NIDCD,“American sign language”, 2017 [Online].

Available:https://www.nidcd.nih.gov/health/american-sign-language, [Accessed April 06,
2019].

3. Suharjito MT, “Sign Language Recognition Application Systems for Deaf-Mute People
A Review Based on Input-Process-Output”, 2017 [Online].
Available:https://www.academia.edu/35314119/Sign_Language_Recognition_Applicatio
n_Systems_for_Deaf-Mute_People_A_Review_Based_on_Input-Process-Output
[Accessed April 06, 2019].

4. Ibrahim, “Sign Language Translation via Image Processing”, [Online].

Available:https://www.kics.edu.pk/project/startup/203 [Accessed April 06, 2019].

5. NAD, “American sign language - community and culture frequently asked questions” ,2017
[Online] Available:https://www.nad.org/resources/American-sign- language/community-and-
culturefrequently-asked-questions/ [Accessed April 06, 2019].

6. Sanil Jain and K.V.Sameer Raja,“Indian Sign Language Character Recognition” , [Online].
Available:https://cse.iitk.ac.in/users/cs365/2015/_submissions/vinsam/report.pdf
[Accessed April 06, 2019]

37
CHAPTER - 7 APPENDIX -

BASE PAPER

38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57