INDUSTRY INTERNSHIP REPORT

An Industry Internship report submitted in partial fulfilment of the

requirements for the award of the degree of
BACHELOR OF TECHNOLOGY
IN
COMPUTER SCIENCE AND ENGINEERING

Submitted by:
Suhani Dahiya ([Link] CSE AIML)

Under the guidance of

Ms. Priya Arora
Assistant Professor, Department of CSE
School of Engineering & Sciences

GD Goenka University, Gurgaon

1
 Declaration by the Student

I hereby grant permission to GD Goenka University to publish my submitted project/research

work for academic purpose. I understand that the published copy may be accessible to the
public through the university's library, digital repositories, and other relevant platforms. This
consent encompasses both print and electronic formats.
By signing this consent form, I give my consent to publish my dissertations as described below:

Student Name: Suhani Dahiya

Student Affiliation: GD Goenka University
Title of work: Intern

Date: 3rd July – 31st August

Place: GD Goenka University
Student Signature:

2
CERTIFICATE OF INTERNSHIP

3
 ACKNOWLEDGEMENT

I would like to express my sincere gratitude to GD Goenka University and I am also thankful
to my faculty supervisor, Ms. Priya Arora for their continuous guidance and support throughout
my internship journey. Their encouragement and academic foundation have been instrumental
in allowing me to apply my learning in a professional environment. Their academic support
helped shape this report, and their insights were vital to my learning experience.
I extend my heartfelt thanks to Yugasa Software Labs Pvt Ltd. for providing me with this
valuable internship opportunity. I am particularly grateful to my industry mentor, Mr. Vivek
Mittal, and the entire team at Yugasa Software Labs Pvt Ltd. for their patience, guidance, and
willingness to share their expertise. Their mentorship and insights have greatly enriched my
professional and personal growth.

4
 ABSTRACT

This report summarizes my internship experience at Yugasa Software Labs Pvt Ltd, where I
worked on the project titled "Image-matching and Digit Prediction (Building Machine
Learning Model) supervised ML Model using Deep Learning." The internship was conducted
under the guidance of Mr. Vivek Mittal from 3rd July 2024 to 31st August 2024.

During my internship, I collaborated closely with the development team and contributed to the
design and implementation of deep learning models for image matching and digit prediction.
My responsibilities included developing machine learning algorithms, preprocessing datasets,
and optimizing model performance using techniques such as convolutional neural networks
(CNNs) and other deep learning methods.

This report highlights the methodologies I employed, challenges I encountered, and the
practical knowledge I gained in applying machine learning concepts in real-world scenarios.
The experience significantly enhanced my technical skills, including proficiency in Python,
TensorFlow, and model evaluation, and has prepared me for future challenges in machine
learning and artificial intelligence.

5
 TABLE OF CONTENT

Company Profile........................................................................................................... 7
Introduction to Assigned Job ....................................................................................... 9
Modular Description of Job ....................................................................................... 17
Detailed Analysis of Individual Modules ................................................................... 18
Project Undertaken .................................................................................................... 18
Conclusions and Future Directions ............................................................................ 19
Bibliography ............................................................................................................... 22
References… ............................................................................................................... 23

6
 COMPANY PROFILE

Company Name: Yugasa Software Labs Pvt Ltd

Location: Gurgaon, Haryana, India
Website: [Link]
Overview:
Yugasa Software Labs Pvt Ltd is a dynamic software development company that offers a wide
range of IT services, including custom software development, mobile application development,
web development, and advanced technology solutions such as Artificial Intelligence (AI) and
Machine Learning (ML). The company is known for its innovative approach in delivering high-
quality, scalable, and reliable solutions to its clients. Yugasa caters to industries ranging from
healthcare and education to retail and finance, providing them with technology-driven solutions
that improve business performance.
Services:
 Custom Software Development: Tailored solutions to meet specific business needs.
 Mobile Application Development: Developing robust mobile apps for Android and
iOS platforms.
 Web Development: Creating high-performance websites and web applications.
 AI and Machine Learning: Specializing in deep learning, neural networks, and natural
language processing (NLP) for complex business challenges.
 Cloud Solutions: Enabling businesses to move to the cloud with secure and scalable
solutions.
Expertise:
Yugasa Software Labs has a highly skilled team of professionals with expertise in various
domains, including:
 Artificial Intelligence & Machine Learning: Implementing cutting-edge AI solutions
to automate tasks, predict outcomes, and enhance decision-making.
 Big Data: Providing analytics and data management services to drive business
intelligence.
 Blockchain: Offering secure and transparent solutions based on blockchain technology.
 Internet of Things (IoT): Developing connected devices for enhanced automation and
control.
Industry Focus:
Yugasa Software Labs serves a diverse set of industries, including:
 Healthcare: Implementing AI for diagnostic tools, patient management, and health
monitoring systems.

7
  Retail: Building solutions for e-commerce, inventory management, and customer
engagement.
 Finance: Providing secure and scalable financial software solutions for banking,
investment, and trading.
 Education: Developing e-learning platforms, educational tools, and student
management systems.
Vision:
Yugasa Software Labs envisions becoming a leading global provider of innovative software
solutions by harnessing the power of emerging technologies to solve real-world problems. The
company aims to transform businesses with customized, high-quality software applications that
drive growth and efficiency.
Mission:
The mission of Yugasa Software Labs is to deliver exceptional software solutions and services
that provide tangible value to clients by focusing on customer satisfaction, quality, and
technology-driven growth. The company strives to empower organizations with the best in
technology while fostering a culture of innovation and collaboration.

8
 INTRODUCTION TO ASSIGNED JOB
The primary objective of my role was to develop a comprehensive system for analyzing
operational efficiency and performance consistency across a series of wells in the oil and gas
sector. This project aimed to provide actionable insights by identifying wells that consistently
operate above or below average, as well as those that could benefit from targeted optimization
measures to enhance efficiency.

This task involved working with two distinct datasets containing operational metrics for
various wells. The project’s key objectives were as follows:
 Data Extraction and Alignment: Standardizing data from multiple sources to enable
consistent analysis.
 Performance Analysis: Identifying wells that operate consistently at higher or lower
levels relative to industry standards.
 Comparative Visualization: Developing visual representations to facilitate clear
comparisons of well performance across parameters.
 Optimization Identification: Establishing a structured framework to pinpoint wells
that could benefit from targeted operational adjustments.
This system was designed to enhance decision-making in well management by providing
insights into performance trends and areas for potential efficiency improvements.

Responsibilities and Contributions

In this role, I was responsible for utilizing advanced machine learning techniques to design and
implement models focused on image matching and digit prediction. These tasks required deep
learning methodologies to ensure high model accuracy and efficiency, alongside data analysis
and feature extraction techniques that contributed to achieving project objectives.

My responsibilities involved extensive collaboration with the development team, where I

applied machine learning algorithms to analyze, align, and interpret complex datasets. Using
Python programming, along with libraries such as pandas and numpy, I conducted data
preprocessing and transformation to prepare datasets for analysis. For visualization, I leveraged
matplotlib to generate comparative graphs and charts, which served to convey insights into
well performance, efficiency, and areas requiring optimization.

This project provided valuable experience in data analysis, machine learning model
development, and performance evaluation, while also refining skills in data visualization and
collaborative problem-solving. The outcomes contributed significantly to establishing a robust
system for ongoing performance monitoring and operational optimization in the well
management sector.

9
 MODULAR DESCRIPTION OF THE JOB
The project was structured into several modules, each addressing specific aspects of well
performance analysis and optimization in the oil and gas sector. This modular approach
allowed for a comprehensive and systematic evaluation, with each module contributing to the
project’s overall objective of identifying performance trends and efficiency opportunities.

Data Extraction and Preprocessing Module : This module focused on the initial phase of the
project—data collection and preparation. It involved importing and cleaning the datasets to
ensure that they were ready for analysis.

Parameter Alignment Module: The parameter alignment module was designed to

synchronize the data from both datasets. It ensured that performance parameters across
different wells and datasets were properly aligned, making it possible to compare them
effectively.

Performance Calculation Module: In this module, performance metrics were calculated for
each well based on their operational data. The objective was to identify trends in well
performance and classify wells according to their performance levels.

Comparative Visualization Module: This module focused on creating visual representations

of the performance data, allowing for easy and effective comparisons between different wells
and datasets. The visuals helped stakeholders quickly interpret the data.

Efficiency Metric Calculation Module: The efficiency metrics module aimed to calculate an
efficiency score for each well, helping to determine which wells were operating optimally and
which required further optimization.

Optimization Identification Framework: This module synthesized the results from the
previous modules and provided actionable recommendations for optimizing well operations. It
helped identify which wells had the most potential for improvement and how they could be
optimized.

10
 DETAILED ANALYSIS OF INDIVIDUAL MODULE

1. Data Extraction and Preprocessing Module

This module served as the foundation of the project, responsible for loading and cleaning the
datasets to ensure consistency and readiness for analysis. The module relied heavily on the
pandas library for data import and preprocessing.

 Data Loading: The pandas.read_csv() function was used to import data files
into DataFrames, which are well-suited for structured, tabular data. This function
ensured that the datasets were loaded with their original formatting and allowed for
easy manipulation.
 Data Cleaning and Handling Missing Values: The module identified missing or
inconsistent values and managed them using [Link]() or
[Link](), depending on the need for continuity or specific data fidelity.
This ensured that the datasets contained uniform, quality data for the remaining
modules.
 Efficiency in Numerical Operations: To handle array-based data more efficiently, the
numpy library was utilized, particularly for replacing null values and performing other
numerical tasks. This setup enabled seamless data preprocessing with consistent and
complete datasets for analysis.

11
2. Parameter Alignment Module

This module aligned the parameters across both datasets to facilitate accurate comparisons.
Given that datasets often contain varying formats for similar information, alignment was
crucial for ensuring consistency.

 Column Parsing and Extraction: The module leveraged pandas to access column
names and split() functions to isolate and restructure them based on well identifiers
and parameter names. By splitting column headers, it effectively identified each well
and its corresponding parameters, ensuring data alignment across both datasets.
 Data Reindexing and Alignment: This streamlined data synchronization, making it
easier to compare wells. By using pandas indexing capabilities, the module could
access and manipulate individual columns, ensuring that data for each well was
accurately aligned.

3. Performance Calculation Module

This module was developed to calculate average performance metrics for each well, based on
various operational parameters. Its purpose was to create a structured framework for evaluating
wells with respect to their peers.

12
  Parameter Calculations: Using [Link](), the module calculated mean values
for each parameter within each well. This function’s vectorized operation allows for
efficient mean calculations across large datasets, making it an ideal choice for
performance analysis.
 Numerical Analysis with numpy: The module also employed numpy to calculate an
overall mean for each well across multiple parameters. [Link]() provided a
computationally efficient means of calculating averages, enabling the identification of
wells that consistently perform above or below average.

4. Comparative Visualization Module

This module was essential for presenting the comparative performance results visually. It
provided stakeholders with a clear view of performance trends and highlighted wells requiring
attention.

13
 Bar Chart Creation Using matplotlib: Utilizing [Link](),
this module produced side-by-side bar charts to display performance data. The precise
positioning of bars was achieved with bar_width and [Link](),
enhancing visual clarity.
 Labeling and Customization: The module used matplotlib functions like
xlabel(), ylabel(), and title() to label the graphs and set legends, making
the data comprehensible at a glance. The module’s use of xticks() provided proper
well names on the x-axis, and the rotation functionality enhanced readability.
 Saving Visuals: By utilizing [Link](), the module allowed results to be
exported as PNG files, making the analysis reproducible and easily shareable.

14
5. Efficiency Metric Calculation Module

This module was designed to evaluate each well’s overall efficiency score, using specific
parameters to determine whether wells were optimally utilized or in need of operational
adjustments.

 Overall Efficiency Calculation: Using both pandas and numpy, the module
calculated the mean efficiency across all parameters for each well. The
[Link]() function efficiently aggregated mean values for a large dataset,
streamlining the process of identifying wells with higher or lower performance levels.
 Categorizing Efficiency Levels: By comparing each well’s efficiency to an overall
mean, the module classified wells as either “most efficient” or “requiring optimization.”
This allowed stakeholders to see not only performance differences but also to make
informed decisions on resource allocation for optimization.

6. Optimization Recommendation Module

This module analyzed wells based on efficiency metrics and identified those that could benefit
from optimization efforts. It also offered specific recommendations for each well group.

 Data Analysis with pandas: Using pandas, this module analyzed well performance
data to generate actionable insights on which wells were underperforming and which
were operating optimally. The .describe() and .sort_values() functions
helped in categorizing and ranking wells for easier assessment.
 Advanced Array Operations with numpy: numpy was essential for calculations
requiring multi-dimensional arrays, such as tracking performance trends across
different parameters for each well. This functionality provided a basis for identifying
long-term trends and assessing the stability of well performance.

15
 Graphical Analysis with matplotlib: For visual representation, this module used
matplotlib to generate comparative charts that highlighted underperforming wells.
The visualization aspects enabled more intuitive decision-making, as stakeholders
could quickly interpret data and take necessary actions.

16
 PROJECT UNDERTAKEN

Project Overview:

The project titled "Image-matching and Digit Prediction" was undertaken during the internship
at Yugasa Software Labs Pvt Ltd. The aim was to develop a supervised Machine Learning
(ML) model utilizing Deep Learning techniques to classify and analyze datasets with well
parameters. This industry-driven project focused on extracting well-specific data, aligning
parameters, and comparing well performance across datasets, ultimately enabling insights into
wells that operate consistently and efficiently or require optimization. The project involved
both data processing and visualization, aimed at aiding real-world applications in resource
management and operational efficiency.

Methodology:-

1. Data Loading and Preprocessing:

o The project began by loading two datasets, each containing performance
parameters of multiple wells. Columns were structured with metadata on each
well's operational metrics.
o Key steps included extraction of well names and their associated parameters
from the dataset column names.
2. Parameter Extraction and Alignment:
o The parameters for each well were systematically extracted and aligned across
datasets. This alignment was crucial to ensure that common parameters were
compared consistently, thus enabling accurate performance assessments.
3. Comparative Analysis:
o The aligned data was used to perform a comparative analysis of mean
performance values for each parameter across wells.
o Additional computations identified wells that consistently operated at higher or
lower levels than others, helping in determining overall efficiency and the need
for optimization.
4. Visualization:
o The project utilized data visualization to present comparative results across
parameters and wells, offering clear insights into operational efficiency and
areas for improvement.

Libraries and Tools:-

I. Pandas:
Purpose: Used for data manipulation and analysis.
Usage in Project: Pandas was essential for loading the datasets, handling complex data
transformations, extracting well parameters, aligning data, and organizing information
for analysis and visualization. It allowed the creation of efficient workflows to process
large datasets and facilitated the extraction of well names and performance parameters.

II. NumPy:
Purpose: Used for numerical operations and handling large datasets.

17
 Usage in Project: NumPy was utilized for computational tasks such as calculating
means, standard deviations, and other statistics. It enabled efficient handling of large
numerical arrays, which was necessary for performance comparison across datasets.

III. Matplotlib:
Purpose: Used for data visualization.
Usage in Project: Matplotlib was the primary tool for visualizing the performance
metrics of various wells. Bar charts and other plots were generated to show comparative
results across parameters, providing insights into well performance and helping in the
identification of consistently high- and low-performing wells. The visualizations made
the data analysis results more accessible and actionable for stakeholders.

IV. Jupyter Notebook:

Purpose: An interactive environment for code execution, documentation, and
visualization.
Usage in Project: The project was documented and executed in Jupyter Notebook,
allowing for step-by-step code development, inline visualizations, and detailed
commentary. It facilitated iterative testing and made it easy to keep records of analyses
and observations in a structured, readable format.

Challenges and Solutions:-

1. Challenge: Extracting and Aligning Well Data

 Description: The datasets contained multiple parameters per well, with varying column
names and structures. Extracting well names and aligning corresponding parameters
across datasets was complex, as column names didn’t follow a consistent naming
convention.
 Solution: Created a custom function to parse and extract well names from column
headers based on predefined patterns. Another function was developed to align
parameters by matching names across datasets, ensuring consistency. This approach
reduced manual intervention and allowed automation of well data extraction and
alignment.

2. Challenge: Handling Missing or Inconsistent Data

 Description: Some wells lacked complete parameter data in one or both datasets,
leading to potential inconsistencies during comparison and visualization.
 Solution: Implemented data validation checks to identify and handle missing or
inconsistent parameters. For wells with incomplete data, the code printed warnings and
excluded those parameters from analysis to avoid skewing results. This ensured that
comparisons were accurate and reliable.

3. Challenge: Performance of Statistical Calculations on Large Datasets

 Description: Calculating metrics such as means and efficiency scores for each well,
and for each parameter, was computationally intensive, especially with large datasets.

18
  Solution: Leveraged NumPy’s optimized array operations and used Pandas efficiently
to minimize computation time. By batching data processing and leveraging vectorized
calculations, the performance was significantly improved, enabling quicker analysis.

4. Challenge: Visualizing Multiple Wells and Parameters Effectively

 Description: Comparing performance across multiple wells and parameters made it

difficult to create clear and insightful visualizations that didn’t overwhelm viewers with
information.
 Solution: Standardized visualizations using grouped bar charts to represent datasets
side-by-side for each well and parameter. Created consistent color schemes and legends
to make comparisons easier to interpret. Automated the process of saving plots for
documentation and future reference.

5. Challenge: Identifying Consistently High or Low-Performing Wells

 Description: The requirement to classify wells as high- or low-performing across

parameters required an efficient way to analyze and group the results.
 Solution: Developed functions that calculated overall mean performance metrics and
compared each well’s score to this benchmark. Wells above the benchmark were
flagged as high-performing, while those below were flagged as requiring optimization.
This method provided a clear, data-driven way to categorize well performance.

Results and Impact:-

1. Accurate Comparison of Well Performance: The project successfully identified wells that
consistently operated at higher or lower levels across parameters in two large datasets. By
comparing the mean performance metrics, wells that were efficiently utilized were
distinguished from those needing optimization. This data-driven insight helps prioritize
maintenance and optimization efforts, potentially leading to improved operational efficiency.

2. Enhanced Decision-Making: By aligning parameters and visualizing performance trends

for each well, the analysis provided valuable insights for stakeholders. The structured, visual
comparisons made it easier for decision-makers to identify underperforming wells and allocate
resources toward improvement. This can significantly reduce costs by focusing on areas with
the highest impact on productivity.

3. Improved Workflow Efficiency: Automating the extraction and alignment of well data
allowed for faster processing of complex datasets. This streamlined workflow saves time for
future analyses, making it easier to replicate the methodology with new data. The efficiency
gained also minimizes the need for manual data cleaning and alignment, allowing engineers
and analysts to focus on strategic tasks.

4. Impact on Future Projects: The modular code structure and robust methodology developed
in this project provide a reusable framework for similar projects in the industry. The success
of this analysis can serve as a benchmark and guide for future data processing and performance
analysis efforts in well management. The approach can be applied to other industry sectors that
rely on similar well or parameter-based data structures.

19
5. Data-Driven Optimization for Well Operations: By identifying wells requiring
optimization, the project allows the company to prioritize these wells in operational strategies.
This proactive approach can lead to improved asset utilization, reduced downtime, and
increased productivity, ultimately enhancing profitability.

Learning Outcomes:

1. Data Processing: Improved data handling skills, particularly with messy, complex datasets,
using pandas.
2. Modular Code: Developed reusable functions for efficient and readable code, useful for
scaling projects.
3. Visualization: Strengthened data visualization skills with matplotlib to communicate
insights clearly.
4. Data Alignment: Learned to align data from multiple sources, crucial for integrating
disparate datasets.
5. Problem-Solving: Tackled real-world challenges, enhancing adaptability and critical
thinking.
6. Statistical Analysis: Gained proficiency in calculating and interpreting efficiency metrics.
7. Real-World Applications: Experienced practical industry applications of machine learning
for resource optimization.
8. Project Management: Enhanced workflow organization and task prioritization.
9. Collaboration: Improved teamwork and communication skills in a professional setting.
10. Python Proficiency: Advanced technical skills with Python and essential data science
libraries (pandas, numpy, matplotlib).

20
 Conclusion and Future Directions

Conclusion:-
The project successfully met its objective of facilitating the analysis and comparison of well
performance across two datasets. Through a systematic approach to data extraction,
alignment, and visualization, the project provided actionable insights into wells that
consistently operated at high or low levels and identified those that were most efficient versus
those requiring optimization. These findings support informed decision-making, leading to
improved resource allocation and operational efficiency in the industry.

Future Directions:-

1. Automating Data Pipelines: Integrate real-time data feeds and automate the data
pipeline to support continuous monitoring and timely updates of well performance
metrics.
2. Implementing Advanced Machine Learning Models: Leverage advanced machine
learning techniques, such as clustering and predictive models, to forecast well
performance and facilitate early anomaly detection.
3. Enhanced Visualization Capabilities: Develop an interactive dashboard to visualize
well performance and efficiency metrics, allowing stakeholders easier access to
insights and trends.
4. Integration of Additional Data Sources: Incorporate external datasets, such as
environmental and operational parameters, to deliver a more comprehensive analysis
of factors influencing well efficiency.
5. Scalability: Adapt the framework to accommodate larger datasets and extend its
application across different sites or resources, enhancing its value for broader
operational analysis.

21
 BIBLIOGRAPHY

I. Python Data Analysis Libraries - McKinney, W. (2017). Python for Data Analysis:
Data Wrangling with Pandas, NumPy, and IPython. O'Reilly Media.
 Provides comprehensive guidance on using pandas and numpy for data
manipulation, essential to the data processing steps of this project.
II. Machine Learning and Efficiency Analysis - Géron, A. (2019). Hands-On Machine
Learning with Scikit-Learn, Keras, and TensorFlow. O'Reilly Media.
 Offers insights on applying machine learning techniques, especially relevant
for developing efficiency metrics and identifying patterns in data.
III. Data Visualization - Hunter, J. D. (2007). Matplotlib: A 2D Graphics Environment.
Computing in Science & Engineering, 9(3), 90-95.
 Introduces matplotlib, the visualization library used for creating
comparative bar charts and other visual analyses in the project.
IV. Industry Background: Oil and Gas Well Operations - Dake, L. P. (2001).
Fundamentals of Reservoir Engineering. Elsevier.
 Provides foundational knowledge on well operations and reservoir
engineering, helping to understand key metrics relevant to well performance.
V. Python Documentation and Tutorials - The Pandas Development Team. (2024).
Pandas Documentation. Retrieved from [Link]
 Official documentation for the pandas library, providing detailed references
on DataFrame functions, data manipulation techniques, and performance
optimization.

The Matplotlib Development Team. (2024). Matplotlib Documentation. Retrieved

from [Link]

 Official documentation for matplotlib, covering the usage of plotting

functions applied in visual analysis.

22
 REFERENCES

1. Van Rossum, G., & Drake, F. L. (2009). Python 3 Reference Manual. Scotts Valley,
CA: CreateSpace.
2. VanderPlas, J. (2016). Python Data Science Handbook: Essential Tools for Working
with Data. O'Reilly Media.
3. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press.

23