
histories to offer personalized advertisements and
product recommendations (Jones, 2019).
In recent years, research on consumer forecasting
has evolved significantly, leveraging advancements
in data analytics, machine learning, and behavioral
economics. These studies focus on predicting
consumer behavior patterns to help businesses make
informed decisions regarding product development,
marketing strategies, and inventory management.
One prominent area of research is the application of
machine learning algorithms to predict consumer
preferences and purchasing behavior. These
algorithms analyze large datasets containing
historical purchase data, social media activity, and
other relevant metrics to identify patterns and trends
(Chen, 2012). By utilizing techniques such as
regression analysis, decision trees, and neural
networks, researchers have been able to improve the
accuracy of their predictions significantly. Another
crucial development is the incorporation of sentiment
analysis in consumer forecasting. Sentiment analysis
involves examining text data from online reviews,
social media posts, and other user-generated content
to gauge public opinion about products and services
(Feldman, 2013). This method has proven effective in
predicting sales and market trends, as it captures real-
time consumer attitudes and sentiments.
Behavioral economics has also contributed to the
advancement of consumer forecasting. Researchers in
this field study how psychological factors and
cognitive biases influence consumer decisions
(Kahneman, 1979). By integrating behavioral
insights with traditional economic models, they have
developed more comprehensive and accurate
forecasting models that account for irrational
consumer behavior. Big data analytics has emerged
as a critical tool in consumer forecasting, allowing
researchers to handle vast amounts of data from
various sources. Integrating big data with advanced
analytics techniques has enabled more precise
segmentation of consumer groups and better
prediction of their future behavior (McAfee, 2012).
This approach helps businesses tailor their strategies
to meet the specific needs and preferences of different
consumer segments. The advent of real-time data
processing and analytics has revolutionized consumer
forecasting. Real-time analytics allows businesses to
monitor consumer behavior as it happens, enabling
them to respond quickly to changing market
conditions and consumer preferences (Gandomi,
2015). This capability is particularly valuable in
industries where consumer trends, such as fashion
and technology, can shift rapidly.
Taken together, these important developments in
recent years have significantly improved the accuracy
and reliability of consumer behavior predictions,
providing businesses with valuable insights to guide
their strategic decisions.
In the wake of the pandemic, there has been a
noticeable shift in consumer habits as the global
economy rebounds. By delving into these changes
using analytical models, sellers can gain valuable
insights to better understand and adapt to evolving
consumer preferences. The pandemic has
significantly impacted the global economy and led to
substantial changes in people's lifestyles and
consumption behaviors. Traditional offline
consumption patterns have given way to online
platforms, resulting in a surge in digital payments and
e-commerce. Concurrently, consumer demand and
preferences for products have also evolved. As
vaccination efforts progress and the pandemic
gradually subsides, there are signs of economic
activities returning to normal. However, it's important
to recognize that the post-pandemic era has reshaped
consumer habits and given rise to new trends and
characteristics. In this context, leveraging model
analysis technology becomes particularly important.
By developing and refining models that capture
consumer behavior, sellers can gain a comprehensive
understanding of these evolving behaviors. In the
next section, the author will introduce the dataset,
including its source, basic description, and the
variables involved. Subsequently, there will be an
overview of the predictive model, covering its
parameters and the evaluation metrics used. The
chapter will then examine the results and offer
insights for future research.
2 DATA AND METHOD
The dataset selected by the authors is from the Kaggle
website and is a newer dataset uploaded by DATA
DIGGERS called E-commerce Sales Data 2023-24.
The E-commerce Sales Data dataset offers a
comprehensive compilation of information about user
profiles, product specifications, and user-product
interactions. This dataset serves as a valuable
resource for gaining insights into customer behavior,
preferences, and purchasing patterns on an e-
commerce platform. The dataset consists of three
sheets. The user sheet contains user profiles,
including details such as user ID, name, age, location,
and other relevant information. It helps in
understanding the demographics and characteristics
of the platform's users. The product sheet offers
ECAI 2024 - International Conference on E-commerce and Artificial Intelligence
110