Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Predicting Profit of a Startup Companies using

Machine Learning Algorithms
1 3
Dr. SK. Mahaboob Basha , R. Manikanta ,
Professor, UG Scholar,
NRI Institute of Technology, Dept. of IT,
A.P, India-521212 NRI Institute of Technology,
A.P, India-521212
2
Ch. Sirisha , 4
UG Scholar, P. Ramyasri
Dept. of IT, UG Scholar,
NRI Institute of Technology, Dept. of IT,
A.P, India-521212 NRI Institute of Technology,
A.P, India-521212

Abstract:- The profit earned by a company for a specific development and marketing you do, then a business can make
period depends on several factors like how plenty of time morethan the predicted profit provided the predicted value is
and money a company spends on research and achievable. Task of profit forecast with machine learning
development marketing and alive more. So forecast the using Python. The dataset of profit prediction includes data
profit of a company for a specific period we need to about the research and development spend, Administration
alternation a machine learning model with a dataset that cost, Marketing Spend, State of operation, and the historical
contains factual data about the profit generated by the profit generated by 50startups. For a particular period, the
company. We can set an achievable goal predicting profit profit earned by a company depends on several factors like
is an important task for every business. For example, if how much time and money a company spends on research
the business allots $500 on marketing, it can’t take a and development marketing and many more. So for a
profit of $20,000. correspondingly, there are lot of other specific period predicting the profit of a companywe need to
components on which the profit of a business depends. A convey a machine learning model with a dataset that contains
company should set a goal that can be accomplished. In old data about the profit accomplished by the company. The
the section below, I will walk you buttoned up the task of task of predicting profit is a main task for every business to
profit prediction with machine learning using Python. set an achievable goal. For example, on marketing the
When there is no automated system there is alwaysa hard business spends $500 on marketing, it can’t assume a profit
in predicting profit as there are more considerations to of $20,000. besides, the profit of a business depends on many
be done. Wisdom of Models is a machine learning other factors. Therefore, a company should set a goal that
algorithm for predicting profit via research and can be achieved. In the given, I'll lead you through the
development costs, administration cost and marketing Python exercise of machine learning profit prediction. The
spend in a business. Wisdom of Models focusses on other major goal of this subsection is to make clear how key
three machine learning algorithms, svr, random forest concepts areused throughout the text. Scale-ups and start-ups
and linear regression to derive a new prediction. Many are occasionally mentioned interchangeably. Both start-ups
companies, institutions, governments, and Private and scale-ups are the focus of this research. To avoid
organisations are funding these businesses and enticing excessive repetition, these may alternatively be referred to as
people to utilise them to investigate their ideas. organisations, groups, realities, and topics of study.
Independent variables are mostly explored as predictors in
Keywords: Research and Development, Wisdom of Models, this study. These can be thought of as component or plain
Administration Cost, Marketing Spend, SVR, Random independent variables to avoid the word "continuous." The
Forest, Linear Regression. dependent variables are mostly discussed as criterion in this
study. These can also be referred to as easily dependent
I. INTRODUCTION variables, success criteria, or estimates of success to avoid
repetition. Success: In this study, success is characterised as
The project titled ‘PROFIT PREDICTION’ it can a binaryvariable. Based on the employed dependent variable,
predicate the profit. The main aim of the project is to predict a start-up is categorised as eithersuccessful or unsuccessful.
the profit. A company should always set a goal that must be
achievable, else, for their potential employees will not be  Problem Statemenet:
able to work if they find that the goal set by the company is For both the administration and the stockholders,
unachievable. profit prediction for a particular period we can predicting the performance of start-up businesses is essential.
set a goal of the profit prediction. If you know how much It is difficult because there aren't enough relevant facts or
profit you can make with the amount of research and reliable generic approaches.[Fig:1] Algorithms that use

IJISRT23MAY429 www.ijisrt.com 201

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
machine learning can predict the future. The future success based on sample data, machine learning algorithms can be
of start-ups must be predicted for both venture capital (VC) construct known as training data, in order to build
investors and startup businesses. Start-up organisations can predictions or resolve without being absolutely programmed
change their development strategy and successfully snare to do so. In a wide variety of applications machine learning
openings by forecasting the future development of algorithms are used, such as in medicine, email filtering,
themselves and their competitors. voice recognition, cultivation, and computer vision, where it
is hard or impractical to develop ordinary algorithms to
perform the require tasks. It is subset it can closely related to
computational statistics, which can focus on making
predictions using computers, but not all machine learning is
numerical learning. The study of mathematical reduction
delivers methods, theory and application region to the range
of machine learning. Data mining is a related plot of study,
through unsupervised learning it can focus on exploratory
analysis. Some fulfillment of machine learning use data and
neural networks in a way that acts the working of a
biological brain. Across business problems its application,
machine learning is also indicate to predictive analytics.
Learning algorithms work on the base that strategies,
algorithms, and consequences that worked well in the
history are likely to continue working well in the
future[Fig:2]. These consequences can be egregious, similar
as "since the sun rose every morning for the last 10,000
days, it will presumably rise hereafter morning as well".
They can be nuanced, similar as "X% of families have
geographically separate species with color variants, so there
is a Y% chance that undiscovered black swans live".

 Linear Regression Model:

In statistics, linear regression is a single approach for
modelling the relationship between a scalar response and
one or more analytical variables (also known as dependent
Fig 1 Startup Companies Profit Prediction and independent variables).[Fig:3] The case of one analytical
variable is called simple linear regression; for more than one,
 Technologies Used: the process is called multiple linear regression. This term is
specified from multivariate linear regression, where multiple
activated dependent variables are predicted,rather than a one
scalar variable.

Fig 2 Technologies Used

 Machine Learning:
Machine learning (ML) is a field of inquiry devoted to
comprehending and creating "learning" methods, i.e.,
methods that use data to enhance performance on a particular
set of tasks. It is a part of artificial intelligence. A model Fig 3 Linear Regression Model in MachineLearning

IJISRT23MAY429 www.ijisrt.com 202

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Random Forest:  Pandas:
A random forest is created by growing many trees. Join Pandas is a Python data analysis library. Wes
However, individual trees are grown using only a subset of McKinney founded pandas in 2008 in response to a demand
the available training data. For each tree, the method chooses for a robust and adaptable tool for quantitative research.
a bootstrap sample of the training data (i.e. randomly chosen Since then, pandas has become one of the most well-known
rows from the dataset). The method then chooses a random Python libraries. It has a very vibrant contributor
sample of the columns (features). The bootstrap sample with community.
a subset of the available features is used to create an
individual tree.[Fig:4] This process is repeated many times  Seaborn:
to grow a “random” forest of trees in which each tree is It is a matplotlib-based Python data visualization
different. 16 When it is time to make predictions, each tree package. It offers an advanced drawing interface for
in the forest is fed the testing data, and each tree makes a beautiful and practical statistics graphics.
prediction. The majority vote of the forecasts made by the
individual trees makes up the prediction of a random forest.  Software Requirement Specification:

 Functional Requirements
 Operating system: Modern operating system (windows
10)
 Coding Language: ML using python
 Front-End: Visual Studio 2012Professional.
 Data Base: SQL Server 2008.
SRS is a captures complete description about how the
system is expected to perform. At the conclusion of the
requirements engineering phase, it is often approved. It
defines how software system will interact with all internal
modules, hardware, communication with each other
programs and human user interactions with a wide range of
real like scenarios.

 Existing System:
By using a single independent variable similar as the
Fig 4 Random Forest in Machine Learning investment cost of a company’s design, the value of the
dependent variable i.e., the profit of the company by the
 Python: means of that design is roughly prognosticated. Linear
It is a well-known high-level programming language that is retrogression makes use of a single independent variable to
employed for general programming. The programming prognosticate the value of a dependent variable by
language was created by Guido van Rossum in the late developing a retrogression line along the given data and
1980s, and it has since become one of the most popular in thereby prognosticating dependent variable using that
use worldwide. For helping developers create a wide range retrogression line. There are some other waysviz the Bracket
of apps, there are several libraries and frameworks available. tree and Random Forest that makes use of a lot of dependent
Because of its well- known straightforwardness and ease of variable to prognosticate the value of the dependent variable
use, it is a good choice for both inexperienced and seasoned and these ways works best for some of the given values
programmers. To expand the power of the language, Python but not for all. The main intention is to prognosticate the
provides a large range of modules that can be employed. value of the dependent variable i.e., the value of the profit of
There are severalcommon modules: the company grounded on the data of the company over the
former times. So, from all the ways used before for the
 NumPy: vaticination of profit an average from all those
NumPy is a general-purpose library for managing prognosticated values of the dependent variable is reckoned
arrays.. Additionally, it provides tools for interacting with and made as the prognosticated dependent variable.
these arrays as well as a high- performance, greater
dimensional array object. In scientific aggregation, Numpy  Drawbacks of Existing System:
is the primary tool for Python.  Unrealistic profit Targets. A cast grounded on
unrealistic hypothetical about the business terrain
 Matplotlib: generally producesunobtainable profit targets.
There are a number of toolkits that are available to  Uses Too important Time
enhance the capability of Python's built-in matplotlib, which  Requires Financial Modelling chops
is not included in the Standard Libraries that are installed by
default. Some of them can be downloaded separately, while
 Results unconnected to Strategies
others can be included in the matplotlib source code  Creates Unrealistic prospects
package but require additional resources.

IJISRT23MAY429 www.ijisrt.com 203

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Linearretrogression makes use of only one i ndependent obtaining an outcome. Everyone is aware that the business
variable and so results are less accurate. sector will offer us options to safeguard our future that will
 Data aren't fully consumed by a direct retrogression change our lives. Our suggested strategy is designed with
model. profit estimation in mind. Because many young stars will
prefer to start their own businesses, we developed our system
 Proposed System: for predicting start-up in this generation. Therefore, we use a
The main intention is to predict the value of the database as an input. The database data includes information
dependent variable i.e., the value of the profit of the company about marketing, administration, research and development,
based on the data of the companyover the previous years. So, and the current state of the company. These prerequisites
from all the techniques used before for the prediction of allow us to generate profit. Users' information can be
profit an average from all those predicted values of the gathered, and our system can then utilise that data to provide
dependent variable is computed and made as the predicted output that is useful to businessesand new startups alike.
dependent variable.
 Advantages of Proposed System:
Training and testing are the two parts of the suggested  It uses all the information provided to it to forecast the
system, which when combined result in a useful tool for any value of the independentvariable.
immovable object. The training phase includes the use of a  Hypothetically it is better than all the other existing
database, data preprocessing, and a machine learning machine learning algorithms.
algorithm classifier. Complete the form to get it ready for
testing. Both the training and testing phases are essential to

 System Architecture:

Fig 5 System Architecture

 Future Scope: model passes the homoscedasticity and no- multicollinearity

The best way to estimate the focus on the certain tests but fails the residual normality test. The model with the
amount of money you are bringing home in a defined lowest RMSE is the forward and backward model, which has
period, not agreement. funds to be honored later on. anadjusted r2 of 0.94. This suggests that thepredictor used in
Ultimately, profit prediction is an in-depth analysis of past the model can interpret 94% of the data, while the remaining
performance to help understand how much your business variables can interpret 6% of the data. We investigate
might bringin during the upcoming year. whether a machine learning (ML) technique called
classification trees can produce out-of-sample profitability
II. CONCLUSION projections that are more accurate than random walk
forecasts. We are driven to usethe ML technique.
This is how machine learning algorithms can be used
to forecast a company's earnings for a specific time frame.
These steps might assist a business in establishing an
achievable goal. According to the assumption checking, the

IJISRT23MAY429 www.ijisrt.com 204

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
REFERENCES BIOGRAPHIES

[1]. Gao, Q., Xu, H., & Li, A. (2022). The analysis of
commodity demand predication in supply chain
network based on particle swarm optimization
algorithm. Journal of Computational and Applied
Mathematics, 400, 113760
[2]. Greg Ross, Sanjiv Das, Daniel Sciro, Hussain Raza,
Capital VX: A machine learning model for startup
selection and exit prediction, The Journal of Finance
and Data Science, Volume 7,2021.
[3]. Daniel Cockayne, what is a startup firm? A Dr. Sk. Mahaboob Basha is presently working as Professor
methodological and epistemological investigation into in the department of Information Technolgy at NRI Institute
research objects in economic geography, Geoforum, of Technology, Vijayawada. He received his M.Tech
Volume 107, 2019. degree from Jawaharlal Nehru Technological University,
[4]. McNamara, Sue; McNamara, Anne (2019). Kakinada (JNTUK) and Ph.D in Computer Science and
Authentic simulated startups: Bringing the real Engineering from Acharya Nagarjuna University (ANU).
world into the classroom. Journal of Education for He has published over 10 research papers in International
Business Conferences and Journals. He has more than 20 years of
[5]. P. Hajek and M. Z. Abedin, "A Profit Function- experience in teaching.
Maximizing Inventory Backorder Prediction System
Using Big Data Analytics," in IEEE Access, vol. 8,
pp. 58982-58994,2020,doi:
10.1109/ACCESS.2020.2983118.
[6]. Florida, Richard; King, Karen M. (2018). Urban
Start-up Districts: Mapping Venture Capital and
Start-up Activity Across ZIP Codes. Economic
Development Quarterly, (), 089124241876373
[7]. Jung, Sang Hoon; Jeong, Yong Jin (2020). Twitter
data analytical methodology development for
prediction of start-up firmsâ social media marketing Ch.sirisha is currently studying B.Tech with specification of
level. Technology inSociety. Information Technology in NRI Institute of Technology. she
[8]. Arroyo, Javier; Corea, Francesco; Jimenez-Diaz, donea summer internship on profit prediction.
Guillermo; Recio-Garcia, Juan A. (2019).
Assessment of Machine Learning Performance for
Decision Support in Venture Capital Investments.
IEEE Access
[9]. Deller Mann, Dominik; Lipusch, Nikolaus; Ebel,
Philipp; Leimeister, Jan Marco (2018). Design
principles for a hybrid intelligence decision support
system for business model validation. Electronic
Markets
[10]. Ramanjaneya Reddy M.” EFFECTIVENESS OF R. manikanta Kumar is currently studying B.Tech with
SOCIAL MEDIA CONNECTED WITH CONTENT specification of Information Technology in NRI Institute of
FOR START UP’S” 2022. Technology. he done a summer internship on profit
[11]. Chih-Fong Tsai., Yu-Chieh Hsiao, “Combining prediction.
multiple feature selection methods for stock
prediction: Union, intersection, and multi-
intersection approaches”, Decision support systems,
50,2010
[12]. Filder.R. Hastings, “The organization and
Improvement of the market Forecasting”, Journal of
operation Research society.45, 1994

P. Ramya Sri is currently studying B.Tech with specification

of Information Technology in NRI Institute of Technology.
she donea summer internship on profit prediction.

IJISRT23MAY429 www.ijisrt.com 205