Traditional e-sales forecasting models for e-commerce platforms face challenges such as handling complex nonlinear data relationships, a lack of personalized marketing strategies, and low utilization of real-time data, resulting in poor forecasting accuracy. To address these issues, this paper explores a machine learning–based approach to sales forecasting, with the aim of improving forecast accuracy and enabling personalized marketing plans. The study uses a public dataset from an e-commerce sales forecasting challenge, performs data preprocessing, and removes outliers and missing values. An e-commerce sales forecasting model is then built using the eXtreme Gradient Boosting algorithm, which can effectively capture nonlinear relationships in the data and generate more accurate sales forecasts. In addition, the K-means clustering algorithm is used to analyze customer data to support the development of personalized marketing strategies.

With the development of economy and the improvement of consumption level, consumers' demand for product style, quality and price is also increasing. The traditional e-commerce model has been unable to meet the personalized needs of users. The Customer to Business (C2B) model is an e-commerce model based on customer' needs, and enterprises can respond quickly. The main purpose of this paper is to discuss how to use the Internet of Things (IoT) technology to analyze and study the C2B supply chain of e-commerce, as well as the strategy of collaborative optimization. For C2B supply chain of e-commerce enterprises (e-CEs), this paper considered the impact of many factors on supply chain pricing, such as Degree of Effort for Personalized Customization (DEPC) r and the impact coefficient of cross price ϕ, which provided a theoretical basis for optimizing pricing decisions.

To solve the problem of low detection accuracy of deep learning targets for crack detection in complex backgrounds, The objective of this study is to propose a fusion (CBAM) attention mechanism to enhance the YOLOv5 model for high-precision bridge crack recognition. At the same time, a consistent bridge crack dataset covering different lighting conditions and humidity environments is constructed. To validate the effectiveness of the proposed model, it was compared and analyzed with current mainstream deep learning models (e.g., CNN, VGG19,etc.).The experimental results indicate that incorporating the CBAM attention mechanism, the proposed model was improved by 5.6% in precision, 1.2% in recall, and 13% in optimization of the integrated discrimination parameters compared to the original YOLOv5 model. In addition, in terms of width detection. The experimental results indicate that the model can reach 0.01 mm in measurement accuracy, The research contributes to the efficiency, precision and intelligence of concrete bridge crack detection.

Based on the LSTM neural network, the author proposes a new data-driven spatial load prediction method to analyze the time series within the neural network, avoid data depression, and determine the correlation of the training data space. Establish prediction models based on different neurons, reduce the dimensionality of collected data through data preprocessing, and ensure data integrity. At the same time, provide data management base, control model input and output, unify process model, ensure training sequence model, combine LSTM neural network model, select prediction method, and finish data-driven related transportation. Experimental results show that the data driven global load forecasting model based on LSTM neural network proposed in this paper takes 1.23 seconds to complete 8000 training data, when traditional data drive spatial load forecasting method based on CNN neural network takes 3.56 seconds to finish 8000 training data. It can be seen that the prediction method proposed in this article has a good prediction accuracy.

Diabetic retinopathy is one of the leading causes of visual loss and with timely diagnosis, this condition can be prevented. This research proposes a transfer learning-based model that is trained using retinal fundus images of patients whose severity is graded by trained ophthalmologists into five different classifications. The research uses transfer learning based on a pre-trained model that is ResNet 50, thus it is possible to train the model with the limited amount of labeled training data. The model has been trained and its accuracy has been analyzed using different metrics namely accuracy score, loss graph and confusion matrix. Such deep learning models need to be transparent for approval by the regulatory authorities for clinical use. The clinical practitioner also needs to have information about the working of the classification method to make sure that he/she understands the decision making process of the model.

The coronavirus pandemic has led to a dramatic increase in depression cases worldwide. Several people are utilizing social media to share their depression or suicidal thoughts. Thus, the major goal of the proposed study is to examine Twitter posts by users and identify features that may indicate depressed symptoms among online users. A numerical metric for each user is proposed based on the sentiment value of their tweets, and it is demonstrated that this feature can detect depression with good accuracy by using several machine learning classifiers. The paper proposes a novel method for measuring the mental health index of an individual by combining the sentiment score with multimodal features extracted from his online activities. A real-time curve is generated using this index that can monitor a person's mental health in real time and offer real-time information about his state. The proposed model shows an accuracy of 89% using SVM, and proper feature selection is very essential for obtaining good performance.