| dc.description.abstract | In the modern era, marketing, which
can be defined as selling and buying, has
expanded in a number of technological fields.
Marketing becomes fruitful when it achieves its
key points, which are called sales and profits. A
most common place to see this selling and buying
process is retailing. Information technology
involves in various marketing fields such as in
prediction processes, data analysis, item
designing and profit calculations. In this study, a
prediction process is primarily developed using
machine learning approaches. Sales item data is
analyzed to predict which items give maximum
or expected profit margins and those which
satisfy the customer the most. There are various
machine learning approaches for aspects such as
sales item prediction, prediction for item features
and item price prediction. The novelty of this
research is that it mainly focuses on special event
items, such as those available in the Christmas
season, items specialized for mothers’ day,
lovers’ day and Vesak festival. The research
process is divided into two main sub-parts; item
classification and item prediction, while both
processes are carried out using several machine
learning approaches. Item classification is done
using four supervised learning classifiers: linear
support vector machine (svc), logistic regression,
multinomial Naïve Bayes, and random forest
classifier. Results prove SVC has maximum
accuracy for classification section, accomplished
using SVC machine learning approach. The
prediction process has been done using the linear
regression approach and according to the
preferred data set, its results prove that database
attribute directly affects the prediction accuracy
and precisions. | en_US |
