Page 186 - Emerging Trends and Innovations in Web-Based Applications and Technologies

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
Figure 6 presents the loss graph of the proposed model. The training loss is initially high, reflecting the model's learning phase.
As the training progresses, the validation loss reduces, signifying that the model is effectively adjusting to the dataset and
improving with each epoch. The reduction in loss as the epochs increase suggests that the model is successfully adapting to the
diverse product categories within Quick Mart's marketplace, thereby optimizing real-time classification for better user
experience and operational efficiency.

Figure 6: Confusion Matrix
The confusion matrix offers valuable insights into the model's ability to classify items accurately across Quick Mart’s platform.
As illustrated in Figure 6, the classifier performs well across 11 product categories, including regular, faulty, refurbished, and
other categories. While the model accurately categorizes most items, a few products from specific categories are misclassified,
which is expected in complex marketplaces with highly varied products. Continuous optimization of the algorithm is required
to further refine the system’s performance. To ensure the system's reliability, key performance metrics, such as accuracy,
precision, and recall, were analyzed for each product category, ensuring that the identification process remains robust and
dependable.

Figure 7: Experimental Results
Figure 7 highlights the improvements in accuracy and the reduction of testing set loss as the number of epochs increases. This
trend signifies that the algorithm is learning from the data over time, leading to more accurate identification and a streamlined
transaction process for second-hand products on the platform. The steady decline in loss and the rise in accuracy emphasize
the system's capacity to adapt to real-world marketplace conditions, ultimately offering users a seamless and reliable
experience when engaging in Quick Mart’s second-hand commerce.
VII. CONCLUSION the system offers an advanced solution for accurately
This work presents a unique and innovative approach to identifying and categorizing items—ranging from normal to
revolutionizing the second-hand commerce industry through refurbished and faulty products—with an impressive
machine learning. By automating the classification and accuracy rate of 92.14%. This high level of precision is
categorization of products within the Quick Mart platform, crucial for the long-term success of Quick Mart, as it

IJTSRD | Special Issue on Emerging Trends and Innovations in Web-Based Applications and Technologies Page 176