www.ijcrt.
org © 2025 IJCRT | Volume 13, Issue 5 May 2025 | ISSN: 2320-2882
Predictive Analytics in E-commerce: Analyzing
Customer Behavior to Enhance Sales Forecasting
1
Faizan Pathan, 2Bhasker Vishwakarma, 3Devendra Kumar Pandey
1
Student, 2Assistant Professor, 3Professor
1
Management
1
Medicaps University, Indore, India
Abstract: The growing complexity and competitiveness of the e-commerce landscape requires such
approaches using advanced data-driven approaches to remain relevant in the marketplace and improve
decision-making. These days, predictive analytics with machine learning and artificial intelligence is going
to be one of the most powerful factors in understanding customer behavior and predicting sales levels more
accurately. Predictive analytic models are changing the game of e-commerce by helping businesses
understand customers, predicting their needs, personalizing their experience and optimizing sales strategies
and this review paper addresses this shift. A review of the literature shows a diverse set of approaches such
as convolutional neural networks, random forests, ensemble learning and hybrid recommender systems.
They are used to process large volumes of structured and unstructured customer data (like browsing habits,
cart behavior, and transactional data) to generate actionable insights. Furthermore, combining big data
analytics with real-time behavioral tracking can dramatically improve the prediction of product demand,
optimize bands abandonment rates, and increase personalized marketing efficiency. Welcome to predictive
analytics, which is indeed promising, but it also faces challenges such as data sparsity, cold start, and the
need for explainable models. This paper also discusses significant opportunities for future research,
including but not limited to multimodal data utilization, improved customer segmentation, and scalable AI
systems that scale and adapt to the dynamic nature of consumer behavior. Through a systematic review of
new developments, case studies, and applications, this study offers an overview of the pivotal role that
predictive analytics is playing in reshaping the future landscape of e-commerce.
Keywords— Predictive analytics, E-commerce, Customer behavior, Machine learning, Sales forecasting,
Artificial intelligence, Data mining
I. INTRODUCTION
The migration of retail transactions to the digital space has resulted in an explosion of online exchanges, and
with them, a new set of challenges and opportunities for players across the e-commerce spectrum. With
online platforms gathering staggering amounts of structured and unstructured customer data, the power to
turn this data into actionable insights has become a competitive imperative. In this context, there is a growing
focus on predictive analytics, which leverages historical data to anticipate future outcomes. Predictive
analytics, a branch of data science, leverages machine learning algorithms, statistical techniques, and real-
time data processing to provide businesses with the insights necessary to make informed decisions regarding
customer acquisition, retention, and revenue optimization [1][2]. It covers diverse fields like marketing,
supply chain, and customer service and provides functions such as demand forecasting, customer lifetime
value prediction, behavior-driven targeting, and more. With complexity and user engagement in e-commerce
using intelligent automation and predictive analytics as a basis of the developed strategic agility [3][4].
The customer behavior analytics has become the bedrock of predictive systems, moving businesses beyond
just knowing what customers buy, and into the realm of why they are, and how they make those decisions.
Most predictive models start with some of the behavioral data which can come in multiple ways like product
views, search queries, dwell time, cart abandonment, purchase frequency etc. These behavioral markers are
IJCRT2505239 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org c120
www.ijcrt.org © 2025 IJCRT | Volume 13, Issue 5 May 2025 | ISSN: 2320-2882
input into algorithms that best identify patterns and correlations from what you did in the past versus what
the likely result of that behavior will be in the future. For example, recommender systems use collaborative
or content-based filtering to suggest products to a user based on their preferences or demographic profile
[5][6]. Moreover, clustering algorithms can be employed to group users into behaviorally similar clusters,
providing for more granular targeting efforts for promotions and engagement. Behavior has an especially
strong correlation to sales, particularly in a real-time system where personalization can directly affect
whether a customer decides to purchase an item. Combining behavioral analytics with sales forecasting
models allow businesses to accurately predict spikes of demand, adjust pricing models accordingly and
reduce inventory risks [7].
This review paper aims to address the role of predictive analytics within the realm of e-commerce,
specifically focusing on the means of understanding customer behavior with the target of improving the
accuracy and strategic benefits of sales forecasting. The range of content also includes a broad walk-through
to predictive methods in general, such as supervised and unsupervised learning, neural networks, and hybrid
recommender systems, illustrating their use case for prominent e-commerce processes. This review is based
on academic studies and applied research, which we will use to examine how these techniques have been
applied to better model customer decision-making and increase conversion rates [8][9]. It concludes by
discussing the increasing reliance on artificial intelligence for the development of adaptive and scalable
models which perform well in high-velocity settings such as online platforms. Additionally, the paper
explores the gaps in the literature at present including model interpretability, data sparsity problems, and
ethical issues about customer profiling. The goal of this paper is to serve as a basis for future research and
practices and help practitioners exploit their data in a more reliable and ethical manner with respect to data-
driven e-commerce strategies, by addressing both technical foundations of predictive analytics and practical
implications for e-commerce data [10][11]
II. TECHNIQUES AND APPROACHES IN PREDICTIVE ANALYTICS
A. Overview of Traditional vs. Modern Predictive Models
Traditionally, e-commerce firms employed traditional predictive models like linear regression, logits
regression, etc. for predicting sales outcomes and customer behaviors. These models were appreciated for
being simple, interpretable and computationally cheap. But this bandwagon of algorithms usually didn’t
stand up to the high-dimensional, noisy or, more importantly, the non-linear datasets ubiquitous in modern
e-commerce’s [1][2]. First, we note that machine learning enables far more complex relationships in the data
to be learned, as these models do not require assumptions about distributions or linearity. Recent methods
including decision trees, ensemble approaches, and deep learning architectures have shown to outperform,
especially when it comes to unstructured data and capturing complex user behavior patterns [3][4]. This
evolution of predictive analytics from the traditional to modern is not simply technological, but it is also a
transformative way of looking at data to create actionable business value in a digital commerce environment.
TABLE 1: COMPARISON OF TRADITIONAL VS. MODERN PREDICTIVE MODELS
Refe
Descript Strength Limitati
Model renc
ion s ons
es
Predicts Poor
Linear based on Simple, with
[1][2
Regres linear interpret non-
]
sion relations able linear
hips data
Forecasts Effective Sensitive
Time
based on for to [3][4
Series
historical seasonal anomalie ]
Models
trends demand s
Robust,
Ensembl
Rando handles Less
e of [7][8
m large interpret
decision ]
Forest feature able
trees
sets
IJCRT2505239 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org c121
www.ijcrt.org © 2025 IJCRT | Volume 13, Issue 5 May 2025 | ISSN: 2320-2882
Gradient
High
XGBo boosting Requires [9][1
accuracy,
ost framewo tuning 0]
efficient
rk
Learns
features
Good for High
from
behavior computat [5][6
CNN structure
trends in ional ]
d
sessions cost
sequence
s
B. Machine Learning Algorithms (e.g., CNN, Random Forest, XGBoost)
Modern machine learning algorithms are at the heart of e-commerce predictive systems. CNNs have been
adapted for other functions, such as extracting features from time-series and structural datasets, which also
can be useful for predicting sales time-series trends and sequential modeling of user behavior [5][6]. Random
Forest is another popular ensemble method that involves creating several decision trees on randomly selected
data subsets and averaging their results to minimize overfitting and increase accuracy. It is especially
beneficial in e-commerce for processing massive amounts of behavioral and transactional data while
reducing overfitting [7][8]. One of the more popular meta-algorithms for structured data is XGBoost, an
improved gradient boosting algorithm. This ability to cope with missing data, weight features properly and
generalize well makes it an often used tool when facing tasks with high stakes like churn prediction,
promotion and impact analysis and monthly guarantees [9][10]. All these algorithms have their own
advantages, and when combined with one another, they can greatly enhance the accuracy and scalability of
your predictive models.
C. Behavioral Data Sources and Feature Engineering Challenges
As users explore products, navigate through web pages and complete or reject transactions, e-commerce
platforms produce an immense amount of behavioral data. These types of back-end data help us create rich
context around what consumers are trying to accomplish in terms of intent and preference, whether they are
searching for products on websites, browsing product pages, or clicking to purchase a product. Nonetheless,
the clear conversion of this raw data into features fit for predictive modeling is quite a task. The effective
engineering of features should reflect both temporal and categorical dimensions of behavior, such as visit
frequency, time between sessions, or interactions of multiple actions within one browsing event [11][12].
Data sparsity is another challenge, particularly for new users or less popular products that have limited
interaction histories, making it difficult to train reliable models. Cold start problems, for instance, must be
addressed through novel approaches like transfer learning or the use of auxiliary information, such as that
from demographic or social media data [13][14]. So, the very powerful predictive capacity that behavioral
data offers requires the use of advanced preprocessing, encoding and imputation techniques in order to
harness its true capabilities.
IJCRT2505239 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org c122
www.ijcrt.org © 2025 IJCRT | Volume 13, Issue 5 May 2025 | ISSN: 2320-2882
TABLE 2: COMMON BEHAVIORAL DATA SOURCES IN E-COMMERCE
Data Predictive Use
Description References
Source Cases
Tracks
Session
Clickstream every page
prediction, [11][12]
Data click by the
funnel analysis
user
Product
Search Logs user
recommendation, [13][14]
Queries searches
intent modeling
Tracks
Abandonment
Cart add/remove
prediction, [6][7]
Behavior actions in
promo targeting
carts
Lifetime value
Purchase Transaction
prediction, [8][9]
History logs
retention scoring
Mentions Sentiment
Social
and analysis, trend [1][15]
Media Tags
hashtags prediction
D. AI-Driven Segmentation and Recommendation Systems
E-commerce companies have embraced artificial intelligence to transform the way they segment their
customers and offer individualized product recommendations. Old models segment consumers by
demographics or fixed preferences, but increasingly we now bring dynamic, AI-driven models that evolve
instantly from behavioural input. Customer segmentation can be achieved using unsupervised learning
techniques, such as k-means clustering or hierarchical clustering, enables businesses to create characteristics
of the data showing natural groupings based on shopping patterns, browsing behavior, purchase history
[15][16]. The segments can be targeted to various types of marketing like discounts, offers, landing pages,
etc., for better engagement and conversion. A third major AI tool are recommendation systems that use
collaborative filtering, content-based filtering, and hybrid models to predict the products that a user will
likely buy. To implement this, such systems can compare similarities among users and products to offer
smart recommendations — leading to higher average order values and better customer retention. Other recent
advances that factor in session-level data, device type and other higher-level information such as inferred
user mood obtained from their interaction patterns IL8 identify appropriate recommendations are context-
aware and deep learning-based recommenders that have shown promising results [17] With the combined
segmentation and recommendation working at the same level representing a unified framework, e-commerce
platforms can customize experiences that work smoothly and are more personalized and relevant to their
customer journey.
Applications in E-commerce Sales Forecasting
E. Case Studies and Implementation Examples from Literature
There are many case studies that highlight the applied element of predictive analytics in improving sales
predictions in an eCommerce setting. One study, for example, used machine learning models to analyze
customer website browsing and purchasing history and attained a substantial accuracy improvement of
forecasts in both the short term and the long term [2]. A distinct instance depicted the organization of an
Egyptian e-commerce company that leveraged techniques from supervised learning, enabling the company
to effectively decide on transactional history data to monitor and manage demand peaks during peak seasons
[4]. These examples show how organizations can convert raw consumer interaction data into substantive
sales forecasts for use in operational planning. It also emphasizes that algorithms must be customized to
particular industrial, geographic and consumer dynamics in order to deliver optimal results — predictive
systems both will and must behave differently across various markets.
IJCRT2505239 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org c123
www.ijcrt.org © 2025 IJCRT | Volume 13, Issue 5 May 2025 | ISSN: 2320-2882
F. Sales Prediction, Inventory Optimization, and Promotion Planning
Sales forecasting models to do not exist in a vacuum, and are inherently linked with inventory control and
promotional planning. By being able to predict demand accurately, it helps avoid the scenario of overstocking
or stockouts, which in turn streamlines the warehouse operations while improving customer satisfaction.
Models based on random forest and XGBoost have already been successfully used to predict product-level
demand allowing proactive inventory replenishment and warehouse space allocation [6][7]. Predictive
analytics also helps with evaluating promotional effectiveness by identifying what marketing action was
associated with a lift in sales and engagement. Retailers can model different promotion scenarios and
expected outcomes and tweak parameters of the campaign accordingly [8]. Therefore, integrated planning
systems have predictive analytics at its core that binds marketing, logistics, and procurement functions into
one seamless process, facilitating a fast-paced e-commerce process.
TABLE 3: USE CASES OF PREDICTIVE ANALYSIS IN E-COMMERCE
Use Case Description Applied References
Model(s)
Sales Forecasting Predicting XGBoost, [9][10]
product- CNN
level future
sales
volumes
Inventory Dynamic Random [6][7]
Optimization stock Forest
management
based on
demand
prediction
Promotion Measuring Regression, [3][4]
Effectiveness impact of Time
campaigns Series
on revenue
and clicks
Churn Prediction Forecasting Decision [13][14]
customer Trees,
attrition Ensemble
Recommendation Personalized Hybrid [15][16]
Systems product Filtering
suggestions
G. Real-time Personalization and Dynamic Pricing Models
Predictive analytics is not only for forecasting but also uses real-time personalization and adaptive pricing
approaches. Tracking user sessions, time spent on products, and interactions with cart on the e-commerce
systems can personalize recommendations and promotional offers that a user may see during their active
visit, which allows for better conversion rates 10. Creating such interactions on the fly is done through deep
learning models such as neural networks and hybrid recommenders which allow you to take into account
changing user tastes in real time. Dynamic pricing models, in contrast, make use of predictive inputs like
demand elasticity, competitor pricing, and inventory levels to algorithmically adjust prices. This keeps
pricing competitive and optimized for profitability and customer interest [12]. Having these real time
capabilities does not only improve user experience but also helps in maximizing revenue and differentiation
of the market in a very competitive market.
IJCRT2505239 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org c124
www.ijcrt.org © 2025 IJCRT | Volume 13, Issue 5 May 2025 | ISSN: 2320-2882
H. Challenges in Scalability, Accuracy, and Cold-Start Problems
However, there are a number of challenges that prevent the widespread adoption of predictive analytics in
e-commerce. Scalability is one major problem, especially when models must manage real-time data
concurrent streams from thousands of users on multiple channels. Features such as high computational costs
and fast model updates require robust infrastructure and algorithm optimization [14]. Similar variation of
accuracy can happen based on quality of the data, model architecture selection, and product being forecasted.
As an illustration, fast-moving consumer goods with stable demand patterns are more predictable than
specialized or seasonal goods. Moreover, cold-start problems are a big problem, as new users and products
with little or no historical data cannot be served effectively by state-of-the-art recommender systems. While
solutions such as using auxiliary data, hybrid models or transfer learning have been proposed to alleviate
this problem, they usually necessitate sophisticated integration and fine-tuning. These challenges need to be
addressed for progress towards more robust and inclusive predictive systems in contemporary e-commerce.
III. FUTURE DIRECTIONS AND RESEARCH OPPORTUNITIES
a) Integration of Multimodal Data (e.g., IoT, Visual Search, Social Web)
Predictive analytics will continue to evolve with the introduction of different and multimodal data streams.
Traditional models are based largely on structured data like purchase history and clickstream logs, whereas
new-age platforms are generating more and more complex data in segments like IoT devices, visual search
tools, voice-based queries, and social conversations. For instance, sentiment analysis of customer sentiments
from product reviews and social platforms can provide great context to purchase habits [1]. IoT-enabled
devices, for example, smart home assistants and wearable tech can also provide real-time usage patterns,
enabling highly experienced marketing and demand forecasting. These advancements in textual
representation, are coupled with new approaches in recommendation systems, which take into consideration
images or videos that customers see and the potential use of multi-modal data, such as textual and visual
cues [3]. Fusion of the heterogeneous type of this data requires the use of advanced data fusion techniques
and sound frameworks to facilitate the effective management of both structured and unstructured content in
a scalable manner.
b) Explainable AI and Ethical Considerations in Predictive Systems
As predictive models in e-commerce grow more complex—particularly with the integration of deep learning
and black-box algorithms—the need for interpretability and ethical transparency becomes critical.
Businesses, consumers, and regulators alike demand to understand how automated decisions are made,
especially when these decisions influence pricing, personalized recommendations, or credit-based eligibility.
Explainable AI (XAI) seeks to address this challenge by offering tools and frameworks that make algorithmic
decisions transparent and understandable to non-technical stakeholders [5][6]. XAI methods such as SHAP
(Shapley Additive Explanations), LIME (Local Interpretable Model-agnostic Explanations), and surrogate
models provide insights into feature importance, enabling analysts to identify which user attributes or
behaviors most influenced a prediction.
Beyond technical transparency, ethical considerations must also be integrated into predictive system design.
Bias in training data can lead to discriminatory outcomes, where specific customer segments receive
systematically different treatment based on non-meritocratic features. For instance, models trained on
historical sales data might inadvertently disadvantage minority groups if past marketing strategies were
skewed or exclusionary. To counteract this, fairness-aware machine learning algorithms are increasingly
being explored to audit and mitigate such biases [7][8]. Moreover, issues such as data privacy, user consent,
and algorithmic accountability are becoming focal points in regulatory frameworks like GDPR in Europe
and CCPA in California.
In the context of e-commerce, ethical AI also involves respecting the boundaries of user autonomy. Hyper-
personalization, if unchecked, may cross into manipulation—nudging users toward behaviors that benefit the
platform more than the customer. Ensuring that recommender systems are designed not only for business
performance but also for user empowerment is a key priority for ethical implementation. As a result, the
future of AI in e-commerce must strike a balance between predictive accuracy and ethical responsibility,
emphasizing interpretability, fairness, and transparency as core design principles.
IJCRT2505239 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org c125
www.ijcrt.org © 2025 IJCRT | Volume 13, Issue 5 May 2025 | ISSN: 2320-2882
c) Enhancing Model Adaptability to Customer Behavior Shifts
Consumer preferences and behaviors are in continuous flux, subject to external stimuli (e.g. economic
climate, macro trends, seasonality, or virality). This implies that predictive systems need to learn
continuously and adapt very quickly to be useful. In such environments, static models make little sense;
demand patterns can evolve rapidly in response to new product categories or sudden demand spikes. Various
approaches, such as online learning [8], reinforcement learning [9], and adaptive ensemble techniques [10],
can also be used to update models progressively when new data becomes available. Another exciting avenue
is the implementation of meta-learning and model retraining pipelines that fine-tune parameters based on
recent engagements with users. Therefore, these advancements can reduce the latency between a change in
behavior and system response, improving the resilience and responsiveness of predictive models in real time
applications [11].
d) Opportunities for Cross-Platform Analytics and Omni-Channel Insights
As consumer journeys increasingly traverse both digital and physical touchpoints, unifying data and being
able instantly to use predictive insights across platforms has become a vital strategic asset in e-commerce.
Customers can look at a product on a mobile app, read reviews on a social media page and complete the
purchase using a desktop site or brick-and-mortar store. From each of these interactions, unique behavioral
data is generated that, when analyzed in isolation, only gives us a small piece of the customer puzzle. The
limitations of the above-stated definitions of user profile data are overcome by using cross-platform analytics,
where user profile data is gained by integrating data from multiple sources that can be used to build user
profiles in real-time [13]. With this 360-degree perspective, companies can provide tailored experiences and
unified messaging across channels.
Omni-channel prediction systems take this one step further, allowing businesses to predict and respond to
customer behavior in a unified way, no matter what device or environment they are in. For instance,
predictive models can provide the right time to send a push notification on mobile or show personalized
banners on the web, depending on individual engagement patterns. Moreover, these cross-device attribution
models help understand the weight of different channels in the funnel before the final purchase is made, so
that marketing spend, and the effectiveness of campaigns can be allocated with more accuracy [15].
In order to fulfil such real time-triggered processes, there are major technical challenges to overcome such
as the alignment of data schemas, resolution of user-identities and real-time synchronization among other
things. Middleware platforms (such as segment) and customer data platforms (CDPs) aggregate and
standardize data from different systems. This fall with them are all the advances in API integration and edge
computing which are allowing for more seamless and scalable cross-platform solutions. In addition, omni-
channel models benefit from predictive power recovery through enhanced data tools to identify micro-
conversions (e.g., product views, add to wish list, review interaction) [17].
As digital ecosystems intertwine even further, the democratized ability to action cross-platform insights will
help dictate a brand’s agility to compete effectively within the era of personalized commerce.” Such research
and innovation in this space
TABLE 4: FUTURE RESEARCH AREAS AND TECHNICAL OPPORTUNITIES
Tools/ Refere
Research Focus Opportunity Description
Approaches nces
Multimodal Data Combine text, images, IoT, and voice data Data fusion, Deep
[1][2]
Integration learning
Explainable AI Interpret decisions made by black-box SHAP, LIME,
[5][6]
models RuleFit
Adaptive Learning Model real-time behavioral shifts Online learning,
[11][12]
Systems Meta-learning
Cross-Platform Unified analytics across web, app, and in- APIs, Middleware
[15][16]
Predictive Engines store touchpoints frameworks
IJCRT2505239 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org c126
www.ijcrt.org © 2025 IJCRT | Volume 13, Issue 5 May 2025 | ISSN: 2320-2882
ACKNOWLEDGMENT
The preferred spelling of the word “acknowledgment” in America is without an “e” after the “g”. Avoid the
stilted expression “one of us (R. B. G.) thanks ...”. Instead, try “R. B. G. thanks...”. Put sponsor
acknowledgments in the unnumbered footnote on the first page.
REFERENCES
[1]. Alojail, M., & Bhatia, S. (2020). A Novel Technique for Behavioral Analytics Using Ensemble Learning
Algorithms in E-Commerce. IEEEAccess,8,150072150080.https://doi.org/10.1109/ACCESS.2020.3016419
[2]. Fabra, J., Álvarez, P., & Ezpeleta, J. (2020). Log-based session profiling and online behavioral prediction in
e-commerce websites. IEEEAccess,8,171834171850.https://doi.org/10.1109/ACCESS.2020.3024649
[3]. Fedirko, O., Zatonatska, T., Wołowiec, T., & Skowron, S. (2021). Data Science and Marketing in E-
Commerce Amid COVID-19 Pandemic. In European Research Studies Journal: Vol. XXIV (Issue 2).
[4]. Jia R., Li R., Yu M., & Wang S. (2017). E-commerce Purchase Prediction Approach By User Behavior
Data. IEEE.
[5]. Malla Reddy, J., & Prasad, S. (2021). THE ROLE OF BUSINESS ANALYTICS IN E-COMMERCE. 6, 2021.
[6]. Manco, G., Masciari, E., & Ras, Z. (2019). Michelangelo Ceci Corrado Loglisci New Frontiers in Mining
Complex Patterns. http://www.springer.com/series/1244
[7]. Morsi, S. (2020). A Predictive Analytics Model for E-commerce Sales Transactions to Support Decision
Making: A Case Study. In International Journal of Computer and Information Technology. www.ijcit.com
[8]. Niu, X., Li, C., & Yu, X. (2017). Predictive Analytics of E-Commerce Search Behavior Predictive Analytics
of E-Commerce Search Behavior for Conversion.
[9]. Pan, H., & Zhou, H. (2020). Study on convolutional neural network and its application in data mining and
sales forecasting for E-commerce. Electronic Commerce Research, 20(2), 297–320.
https://doi.org/10.1007/s10660-020-09409-0
[10]. Rathod, J. (2021). Analyzing the impact of big data and business analytics in enhancing demand-
driven forecasting in retail. https://www.researchgate.net/publication/378469130
[11]. Saleem, H., bin Muhammad, K., Hussain Nizamani, A., & Saleem, S. (2021). DATA SCIENCE AND
MACHINE LEARNING APPROACH TO IMPROVE E-COMMERCE SALES PERFORMANCE ON
SOCIAL WEB. International Journal of Advanced Research in Engineering and Technology (IJARET),
12(4), 12–16. https://doi.org/10.34218/IJARET.12.4.2021.040
[12]. Saleem, H., Khawaja, M., Uddin, S., Habib-Ur-Rehman, S., Saleem, S., & Aslam, A. M. (2019).
Strategic Data Driven Approach to Improve Conversion Rates and Sales Performance of E-Commerce
Websites. International Journal of Scientific & Engineering Research, 10(4). http://www.ijser.org
[13]. Sarisa, M., Boddapati, V. N., Patra, G. K., Kuraku, C., Konkimalla, S., & Rajaram, S. K. (2024). An
Effective Predicting E-Commerce Sales & Management System Based on Machine Learning
Methods. Journal of Artificial Intelligence and Big Data, 1(1), 75–85.
https://doi.org/10.31586/jaibd.2020.1110
[14]. Usmani Z.A., & Manchekar S. (2017). A Predictive Approach for improving the sales of products in
E-commerce. IEEE.
[15]. Zhang, Y. (2021). Sales Forecasting of Promotion Activities Based on the Cross-Industry Standard
Process for Data Mining of E-commerce Promotional Information and Support Vector Regression. Journal
of Computers,32(1),212225.https://doi.org/10.3966/199115992021023201018
[16]. Zhao, K., & Wang, C. (2017). Sales Forecast in E-commerce using Convolutional Neural Network.
http://arxiv.org/abs/1708.07946
[17]. Zulaikha, S., Mohamed, H., Kurniawati, M., Rusgianto, S., & Rusmita, S. A. (2020). Customer
Predictive Analytics Using Artificial
Intelligence.SingaporeEconomicReview.https://doi.org/10.1142/S0217590820480021
IJCRT2505239 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org c127
