Volume 8, Issue 4, April – 2023 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Exploring Future Sales Prediction using

Classification and Regression
Dr. Chaitanya Kishore Reddy.M1 G.Sravanth 2
Professor & Dean-IT, NRI Institute of Technology, A.P, UG Scholar, Dept. Of IT, NRI Institute of Technology, A.P,
India-521212 India-521212

P.Mounika3
UG Scholar, Dept. Of IT, NRI Institute of Technology, A.P, India-521212

Abstract:- Future forecast The greatest way to If we are trending below target, it offers us time to
accomplish the targeted marketing objectives is through adjust the sales plan, allowing us to manage our revenue
sales analysis. It would be better to advance your career projections for the predicted timeframe.
once you have the capacity to make strategic decisions in
sales forecasting. When predicting sales numbers, it is It gives individual sales reps visibility into their
important to remember that future product prices will ongoing deals, enabling them to anticipate future roadblocks
have an impact on the volume of sales in addition to past and make necessary course corrections.
sales data. Multivariate time series first appear to be the
best model for this problem. Since there is only ever one It makes it possible for the entire organisation to plan
price for a product at any given time in sales history, forward, looking at potential growth prospects in new use
unlike in real life where history is not always repeating. cases or even emerging markets.
It makes creating a multivariate time series more
challenging. However, the price is more dependent on The act of assessing future sales by foreseeing the
the expiration date for some seasonal or perishable volume of goods or services that a single salesperson, a sales
products. This additional data can aid in the creation of team, or an organisation is likely to sell over the course of a
a causal time series model that is more precise. The specific time period, such as the upcoming week, month,
proposed remedy makes use of a univariate time series quarter, or year is known as sales prediction, sometimes
model, but includes the product's price as a factor that known as sales forecasting. There are many different
systematically affects the prediction. Based on previous approaches to estimating sales, but what's crucial is that they
sales data, the price influence is calculated using data are all supported by data of some form. You may replace gut
correlation analysis and customizable price ranges to instincts with evidence-based decision-making by using
find products with comparable histories. This unique sales prediction. There are two ways to forecast sales:
strategy is simple to compute compared to other
methods and enables the pricing parameter for  Since quantitative forecasting makes use of both recent
simulations and forecasts to be chosen in advance. trends and historical company data, it is the most
precise.
Keywords:- Sales Predicton,Forecasting, Linear
Regression,Random Forest,Classification, Regression,  Qualitative forecasting is individualised and draws on
Decision Trees, Training set. expert opinions, market research, and surveys of future
clients.
I. INTRODUCTION
In order to analyse previous company data and
Future sales prediction is the practise of estimating a patterns, such as completed and won deals and win/loss
company's short- or long-term sales success in the future records, sales forecasting software uses mathematical
utilising historical data and sales records from previous methodologies. This analysis results in an accurate report of
years. One benefit of careful financial preparation is this. A predicted sales income. Forecast reports contrast the sales
sales forecast is an estimate of future revenue for a given goals with the actual results and the projected results.
time frame, such as a week, month, quarter, or year.

The forecast is typically based on information like the

amount of deals that have already closed in your sales
pipeline and the stage of the sales cycle they are in.

It enables a sales force to forecast which deals will

complete and how much money will flow into the company
as a result with a high degree of accuracy.

IJISRT23APR1541 www.ijisrt.com 1237

Volume 8, Issue 4, April – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

Fig 1 Sales Forecasting

II. TECHNOLOGIES USED

Fig 3 Machine Learning

 Linear Regression
A machine learning approach called linear regression
is based on supervised learning. The regression task is
carried out. Regression uses independent variables to model
the desired predicted value.

Additionally, it is employed to determine the

connection between factors and predicting.

Regression models vary depending on the amount of

independent variables employed, the type of relationship
among dependent and independent variables, and other
factors.

The dependant variable in a regression goes by many

different names. It may also be referred to as a regressand,
Fig 2 Technologies used endogenous variable, criteria variable, or outcome variable.

 Machine Learning The independent variables may also be referred to as

Machine learning is a subfield of artificial intelligence predictors, exogenous variables, or regressors.[Fig:4]
(AI) and computational science that primarily concentrates
on using data and algorithms to mimic how humans learn, The task of predicting a dependent variable's value (y)
progressively increasing its accuracy. It is one of the crucial with an inverse relationship involving x (input) and y
components of the expanding discipline of data science. (output) is carried out using linear regression.

To produce classifications or predictions or to find Hence, linear regression is so named. In the diagram
significant trends in data mining projects, algorithms are above, a person's pay is represented by Y (output) and their
developed and refined using statistical approaches. Ideally, job experience by X (input). The line that fits a model the
these insights have an impact on important growth metrics best is the regression line.
since they are used to drive decision-making within services
and enterprises.

The market will increasingly want more data scientists

as big data continues to expand. These are necessary to
assist in determining the most appropriate business inquiries
and the data required to respond to them.[Fig:3]

IJISRT23APR1541 www.ijisrt.com 1238

Volume 8, Issue 4, April – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Python
You may create and share documents with live code,
equations, visuals, and narrative text using the free and
open-source Jupyter Notebook web tool. Included in this are
data transformation, classification, statistical modelling, data
visualisation, machine learning, and data cleaning. Python is
one of the nearly 40 programming languages supported by
Jupyter. The Jupyter Notebook software itself can be
installed using Python.

Using Anaconda, install Jupyter: Use the Anaconda

Distribution to install Python and Jupyter, which also
contains additional frequently used tools for scientific data
and computational science. The most recent version of
Python 3 for Anaconda is available here for download.
Install Anaconda as soon as it has been downloaded.[Fig:6]

Using PIP, install Jupyter Notebook by running

python3 -m python install --update pip. install Jupyter using
Fig 4 Linear Regression python3 -m pip
 Random Forest Jupyter Notebook is the command used to launch it.
The grouping technique known as Random Forest uses
several decision trees with a method known as Bootstrap as
well as Aggregation, often known as bagging, to accomplish
both classification and regression analyses.

The primary idea underlying this is to merge many

decision trees in selecting the final output instead of
depending on individual choices trees.

As the primary learning models, Random Forest uses a

variety of decision trees.

For each model, sample datasets are created by

randomly selecting rows and features from the dataset.

The Bootstrap component is this.[Fig:5]

Fig 6 Python

III. SOFTWARE REQUIREMENTS

SPECIFICATION

SRS is a comprehensive description of the system's

projected performance. At the conclusion of the
requirements engineering phase, it is often
approved[Fig:7].It outlines how software systems would
interact with each other, internal components, hardware,
other programmes for communication, and human
interactions with them in a variety of situations that are
similar to real-world ones.[Fig:7]

Fig 5 Random Forest

IJISRT23APR1541 www.ijisrt.com 1239

Volume 8, Issue 4, April – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Inaccuracy
The majority of conventional forecasts assign a weight
to each step of a negotiation, and these weights get better as
the deal approaches a win or loss. They do not, however,
take into account the length of time the arrangement has
been in the works, the level of interaction between the
representatives and the stakeholders, the recentness of the
participation, if the close date remains the same or changed,
and so on. As a result, because it does not take into account
these decisive aspects, this kind of forecast is not accurate.

 Lack of Sales History

Newer companies or startup companies may find it
difficult to estimate sales because sales forecasting models
frequently use prior data to project future sales. Some
methods need at least two years' worth of historical data to
produce a reliable forecast.

Fig 7 SRS V. PROPOSED SYSTEM

IV. EXISTING SYSTEM To enhance the accuracy of predictions, we suggest

using both a random forest approach and a linear regression
To put it simply, "future sales prediction" is a method technique.
used to categorise or forecast future sales for a retail outlet
or other business. This method's major objective is to The relationship among a scalar answer and one or a
precisely predict future sales in order to prevent product few explanatory factors can be modelled linearly using the
loss. Through the use of algorithms for machine learning linear regression method. By using the input variable as a
and models, future sales can be predicted. basis, this method forecasts the output variable. It's a
dependent variable that you're after. Independent variables
Sales forecasting is done utilising information about are those that you use to forecast the value of another
previous performance. Management of the sales pipeline variable.
and forecasting go hand in hand. Sales representatives and
managers can give information for a trustworthy sales When solving classification and regression issues, the
forecast by getting a precise picture of success, dedication random forest algorithm, a sort of supervised learning
and velocity for each contract. a projection of sales provides algorithm, employs ensemble techniques. During training,
a precise forecast of future revenue performance by utilising the algorithm builds a large number of decision trees and
a multitude of data sources. outputs the mean as well as the mode of each tree's forecast.

Typically, historical performance information is used  Advantages of Proposed System

to construct sales projections. A detailed report that
forecasts the amount of sales that an agent, team, or  Advantages of Linear Regression
organisation will make each week, month, quarter, or year is
known as a sales forecast.  Simple to Implement
It's fairly simple to implement linear regression and
Historical prediction: This method for predicting get good results. In addition, compared to other complicated
makes use of current or appropriate data. techniques, these models may be trained quickly and
efficiently even on platforms with limited computational
Forecasting of opportunity stages stage forecasting: capability.Comparing linear regression with alternative
Your sales funnel is related to this forecasting technique. machine learning techniques, linear prediction has a
significantly reduced temporal complexity.The linear
 Disadvantages of Existed System regression's mathematical formulae are also rather simple to
comprehend and interpret. Given this, the linear regression
 Uncertainty method is a pretty simple algorithm to use.
Nothing can be guaranteed to occur in the future.
Global pandemics, economic shifts, and competitive  Results on datasets with linear separation
behaviour are a few examples of variables that are simply In order to determine the kind of a connection between
impossible to forecast. Your forecasts may out to be variables, linear regression frequently used to arrange
meaningless due to outside influences. linearly separable datasets almost flawlessly.

IJISRT23APR1541 www.ijisrt.com 1240

Volume 8, Issue 4, April – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Advantages of Random Forest  Sales Forecasting
Sales forecasting is the procedure of assessing
 In general, random forests, or neural networks, calculate upcoming sales by estimating the volume of goods or
the relevance of variables. Additionally, they provide a services that one salesperson, a team of salespeople, or an
better way of handling missing data. The variable that is organisation is expected to sell over the course of a specific
present in a given node the most often is used to fill in time period, such as a week, every month, period, or
any missing values. annual.[Fig:8]
 Out of all the categorization techniques that are
currently available, random forests offer the highest  Sales Forecasting Results Evaluation
accuracy. When using forecasting techniques, it is important to
 It can also manage enormous amounts of data with estimate the circumstances. To test methods, evaluation
thousands of different variables. When a class in the procedures are required.[Fig:8]
data is more rare than the other classes, it can
effectively balance the data sets. VII. FUTURE SCOPE
 Additionally, this strategy manages variables the
quickest, making it appropriate for challenging tasks. A forecast of sales to come within a specific time
frame, typically for one month or a year, is known as a sales
VI. SYSTEM ARCHITECTURE prediction. The projection is based on information like the
amount of commitments presently in the channel of sales
and where in the sales process each of them are. Forecasting
sales offers several wonderful advantages. The ability to
effectively allocate resources for future growth and manage
product loss is made possible by a company's ability to
predict sales accurately. Sales projections empower
executives to adjust their course early and help set criteria
for future direction. Sales managers can stop merchandise
from being destroyed by using sales prediction. Quotes for
sales and revenue expectations can be aligned, allowing
revenue leaders to maximise their chances of success.

VIII. CONCLUSION

Sales forecasting is an essential step in the strategy

planning process because it enables businesses to foresee
future performance. It gives them the ability to foresee not
only fresh opportunities but also potentially detrimental
tendencies. A mission statement is essential for any
organisation since it defines its goals and serves as a guide
for decisions. Businesses may effectively assess their costs
and revenues with the use of sales forecasting, which then
allows them to project their performance over the short- and
long-term.

Fig 8 System Architecture REFERENCES

 Gathering Data [1]. https://www.geeksforgeeks.org/mllinearregression/#:

In order to answer particular inquiries, investigate ~:text=Linear%20Regression%20is%20a%20machin
hypotheses, and assess results, gathering data is the act of e,relationship%20between%20variables%20and%20f
obtaining and analyzing information on relevant variables in orecasting.
a systematic manner.[Fig:8] [2]. https://spotio.com/blog/best-sales-forecasting-
software/
 Analysis of Data [3]. https://www.tandfonline.com/doi/full/10.1080/21642
Data is used in predictive analysis to foretell future 583.2014.999389?scroll=top&needAccess=tr
patterns or events. Additionally, it uses previous data to [4]. https://corporatefinanceinstitute.com/resources/data-
project prospective outcomes that motivate us to make science/random-forest/
strategic decisions. [Fig:8] [5]. https://iq.opengenus.org/advantages-and-
disadvantages-of-linear-regression/
 Choosing Best Sales Forecasting Model
The top sales forecasting model techniques, such as
"Regression" and "Classification," are picked in this
module. [Fig:8]

IJISRT23APR1541 www.ijisrt.com 1241

Volume 8, Issue 4, April – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
BIOGRAPHIES

Dr. Chaitanya Kishore Reddy. M is currently working as a

Professor and Dean in the Department of Information
technology at NRI Institute Of technology, Pothavarappadu,
Agiripalli, Krishna(dist), India. He received Ph.D. in
Computer Science and Engineering and M. Tech in
Computer Science and Engineering at Jawaharlal Nehru
Technological university, Kakinada. He has Published 40
research papers in various National and International
Journals and International Conferences. He is a member in
ISTE, CSI, and IAENG. His research areas are Mobile Ad-
hoc Networks, IoT, and Cloud Computing.

G.Sravanthi is currently studying B.Tech with specification

of Information Technology in NRI Institute of Technology.
She has done a project on future sales prediction. She has
finished an internship at Blackbucks and she has done 2
NPTEL certificates.

P.Mounika is currently studying B.Tech with specification

of Information Technology in NRI Institute of Technology.
She has done a project on future sales prediction. She has
finished an internship at Blackbucks and she has done 2
NPTEL certificates.

IJISRT23APR1541 www.ijisrt.com 1242