Scribd
Upload
14 views13 pages

Phase 1

Uploaded by

poongavanampalani003
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
14 views13 pages

Phase 1

Uploaded by

poongavanampalani003
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/ 13

A study on the

Earthquake prediction model Using python

Project Report

Submitted by
POONGAVANAM P
Reg no: 513421104028

BACHELOR OF ENGINEERING
In

DEPARTMENT OF COMPUTER SCIENCE

UNIVERSITY COLLEGE OF ENGINEERING KANCHIPURAM


KANCHIPURAM-631532
September,2022
Earthquake Prediction Model
using Python
PHASE-1
Abstract: An earthquake is a type of natural disaster that is well-known for the
devastation it causes to both naturally existing and artificial structures,
including buildings, bungalows, and residential areas, to name a few.
Seismometers, which pick up vibrations caused by seismic waves moving
through the earth's crust, are used to measure earthquakes.

The damage caused by an earthquake was categorised in this work into


damage ratings, which have values ranging from one to five. The damage
grade of a certain structure, which is linked to a Unique Identification String,
was predicted using a previously gathered data set and a number of criteria.
An analysis of current machine learning classifier techniques was used to
make the forecast.

Logistic Regression, Support Vector Machine (SVM), Random Forest


Classifier, and K-Nearest Neighbours were the machine learning techniques
employed in this study. The best algorithm was taken into consideration after a
review of a number of attributes. The method used to predict the property
underwent a thorough investigation, and the data analysis that followed
revealed information that could help future earthquakes' effects be lessened.

Keywords: Machine learning, Support Vector Machine (SVM), Random


Forest Classifier, Logistic Regression, K Nearest Neighbors, and predictive
analysis.

I. INTRODUCTION

A catastrophic event such as an earthquake is harmful to human interests and


has negative effects on the environment. Incalculable harm to buildings and
other assets has always been done by earthquakes, which have also claimed
millions of lives around the world. Numerous national, international, and
transnational organisations implement various disaster warning and
preventive strategies to lessen the effects of such an incident.
Organisation managers have a number of challenges when it comes to
allocating the organisation's resources because time and quantity are
constraints.To estimate the extent of damage done to buildings after an
earthquake, it is possible to use machine learning. This is accomplished by
categorising these buildings according to a degree of damage severity based
on a number of elements, including their age, foundation, number of floors,
kind of material used, and others. Then, ward-by-ward in a district, the number
of families and the likely casualties are considered. This enables the
proportionate distribution of relief forces by ward and their prioritising
according to the severity of the damage.

Such models can contribute to the fastest possible lifesaving and prove to be
a successful and affordable option.[1-3] It can be further enhanced by
including the distribution of goods like food, clothing, medical care, and money
in accordance with the number of fatalities among people and the degree of
structural damage.

II. BACKGROUND

From its foundations in databases, statistics, applied science, theory, and


algorithms, the discipline of machine learning has developed into a core set of
approaches that are applied to a variety of issues. Over the past 20 years,
significant advancements have been made in the scientific and technical fields
of computational modelling and data gathering.

Additional data repositories have been produced as a result of a combination


of sophisticated algorithms, exponentially rising processing power, and precise
sensing and measurement tools. Networks with cutting-edge technologies
have made it possible to send enormous amounts of data around the globe.
This leads to an extreme need for tools and technologies in order to analyse
scientific data sets effectively with the aim of deciphering the underlying
physical events. Machine Learning applications in geology and geophysics
have achieved significant success within the areas as weather prediction,
mineral prospecting, ecology, modelling etc and eventually predicting the
earthquakes from satellite maps.
An interesting aspect of the numerous of these applications is that they
combine both spatial and temporal aspects within the info and within the
phenomena that’s being mined. Investigations on earthquake predictions
support the concept that each one amongst the regional factors is filtered out
and general information about the earthquake precursory patterns is
extracted. Feature extraction involves a pre-selection process of varied
statistical properties of data and generation of a group of seismic parameters,
which correspond to linearly independent coordinators within the feature
space. The seismic within the sort of statistic are often analysed by using
various pattern recognition techniques. Statistical or pattern recognition
methodology usually performs this extraction process. This gives an idea
about mining the scientific data.

III. MOTIVATION

Earthquakes are one amongst the foremost destructive natural disasters.


They typically occur without notice and do not allow much time for people to
react. Therefore, earthquakes can cause serious injuries and loss of life and
destroy tremendous buildings and infrastructures, resulting in great economic
loss. The prediction of earthquakes is clearly critical to the protection of our
society, but it’s proven to be a Challenging task to predict beforehand and yet
we discover this as a motivating problem to be solved.

IV. SOFTWARE REQUIREMENT SPECIFICATION


i. Anaconda Navigator 2.3.2
ii. Jupyter Notebook 6.0.3
iii. Tkinter
iv. OpenCV
v. Pycharm 2022.2.4
vi. Language: Python 3.7
vii. Environment: Keras and Tensorflow environment
viii. OS: Windows 7 or higher

V. METHODOLOGY
Dataset Link:
https://www.kaggle.com/datasets/usgs/earthquake-database
DESIGN THINKING
1.Data Source: Choose a suitable Kaggle dataset containing earthquake data
with features like date, time, latitude, longitude, depth, and magnitude.

A. Importing Libraries :

Figure 1 shows the Python code to import libraries. We have used four
libraries :-
• Python has a library called Numpy that is used for scientific computing. This
library is utilised throughout the project and is imported as np.
• Pandas are used for data analysis and manipulation. An open source,
BSD-licensed library called pandas offers simple data structures and tools for
data analysis. It is imported as pd.
• matplotlib is a python library. The command-style utilities in pyplot enable
matplotlib to behave similarly to MATLAB. It is imported as plt.
• Seaborn is a matplotlib-based Python data visualisation package for
aesthetically pleasing and educational statistical visuals. It is imported as sns.

B. Importing data :

The below figure displays the Python code for importing data from the
appropriate directory or file and allocating it to a DataFrame. It imports the
data that is kept in CSV format.
C. Checking for NaN :

Checking for NaN is a critical step in the pre-processing of data. We were only
able to identify a few NaNs in this test. The Python code to check for NaN is
displayed below .

D. Manipulating NaN values :

It is essential to remove the NaN values. This can be done by


• Removing the entire column containing many NaN values
• Forward fillna method
• Backward fillna method
• Mean method

Here, the data is random; we need to scale according to inputs to the model.
In this, we convert given Date and Time to Unix time which is in seconds and
a numeral. This can be easily used as input for the network we built.
2.Feature Exploration: Analyse and understand the distribution, correlations,
and characteristics of the key features.

Distribution :
Correlation :

Characteristics:

Earthquake prediction typically requires the information regarding the


magnitude, location, and time of occurrence.
These can be categorised into short-term and long-term processes.

3.Visualization: Create a world map visualisation to display earthquake


frequency distribution.

Here, all the earthquakes from the database are visualised on to the world
map which shows clear representation of the locations where frequency of the
earthquake will be more.

E. Plotting a Basemap :-

Basemaps serve as a reference map on which you overlay data from layers
and visualise geographic information. An individual basemap can be made of
multiple feature, raster, or web layers. Basemaps are the foundation for your
maps and provide context for your work. A sample 3D basemap over Athens,
Greece, is shown.
F.PLOTTING A HEATMAP:-

A heatmap is used to assess the correlation between the fields of the


collecting data.When developing various AI prediction models,the magnitude
of the values along with the sign is crucial.

Heatmaps are used in various forms of analytics but are most commonly used
to show user behavior on specific web pages or webpage templates.
Heatmaps can be used to show where users have clicked on a page, how far
they have scrolled down a page, or used to display the results of eye-tracking
tests.
G. Train/Test split :-

Creating train and test sets from the data is our next step towards developing
a Machine Learning model. The Python code to divide the data set into train
and test data is shown below
5.Model Development:

Build a neural network model for earthquake magnitude prediction.


In the above case it was more kind of linear regressor where the predicted
values are not as expected. So, Now, we build the neural network to fit the
data for training set. Neural Network consists of three Dense layer with each
16, 16, 2 nodes and relu, relu and softmax as activation function.

SOLUTION ARCHITECTURE
● Solution architecture is a structured approach to designing complex
systems or projects,
● Outlining the components , relationships , and processes to achieve
specific goals or solve problems efficiently .
● It provides a high - level blueprint for project development and
implementation .
● Gather and preprocess data from various sources , ensuring
data quality .
● Train AI models for earthquake prediction and evaluate their
performance.
● Deploy models in a secure , scalable environment with a user - friendly
interface .
● Continuously monitor and update the system while complying with
regulations .
● Collaborate with professionals and gather user feedback for
improvements .
AI based earthquake prediction system solution architecture:
Conclusion:

In this Machine Learning Project, we explore an earthquake


prediction system. In this model , we used some of design thinking process
such as data source,exploration of dataset features and also visualise the
earthquake frequency using visualisation tools like numpy , pandas , seaborn ,
matplotlib ,etc.,

References :

1. https://github.com/akash-r34/Earthquake-prediction-using-Machine-lear
ning-models
2. https://www.researchgate.net/publication/313858017_Machine_Learnin
g_Predicts_Laboratory_Earthquakes
3. http://earthquake.usgs.gov/data/comcat/data-eventterms.php#time
4. DriveData, Richter’s Predictor: Modeling Earthquake Damage (2021).
5. Sameer, Earthquake History (1965–2016): Data Visualization and Model
Development.
6. World Health Organization, Earthquakes, Health topics.

You might also like