Intern(Chow Jun Hoe)

 30/6/2025 Today, we will continue to study and understand the XGBoost (Extreme Gradient Boosting) model, which is one of the advanced prediction methods used to improve the accuracy of sales forecasting. It is based on the structure of gradient boosting, which minimizes the error by continuously building decision trees.

 27/6/2025 Public holiday

 26/6/2025 Today, I formally submitted the completed analysis task report to Supply Chain Executive, which is the improved Reserve and In Transit data analysis results. In the afternoon, I continued to participate in the meeting on Output (MinMax).

 25/6/2025 After reviewing the analysis of the Reserve and In Transit issues completed previously, we optimized the code and performed almost the same data analysis. Continue yesterday's task and search and extract Raw Remark for the remaining SKUs.

 24/6/2025 Today I completed the task assigned by the Manager. I successfully found the Raw Remark data corresponding to the specified 10 items and organized them into a clear report format for subsequent analysis and follow-up.

 23/6/2025 Today I accepted a new task from Supply Chain Executive and started to analyze the relevant data about Reserve and In Transit Issue to help identify possible problems in inventory flow.

20/6/2025 Today, I successfully completed the construction and comparative analysis of the ARIMA and SARIMA time series forecasting models, and wrote a summary report, which comprehensively sorted out the model development process, parameter selection logic, forecasting effects (such as MAPE), and the differences in applicability of the two models in different sales scenarios. After completing the summary, I reported to the Manager and Executive, clearly demonstrating the principles, execution process and business application value of the forecasting model, and recorded the direction of subsequent optimization based on their feedback.

 19/6/2025 Today, i continue to gain an in-depth understanding of the parameter settings and meanings of the SARIMA model, focusing on the impact of all parameters in the model on the prediction results. I also continuously optimize the parameter combination through actual tests to further improve the prediction accuracy.

 18/6/2025 Today, I successfully completed the analysis task assigned by the Executive. The main content was to count the Out of Stock (OOS) of all products within a specified time range and calculate how many days each SKU was in the OOS state. In the afternoon, I attended a meeting on MinMax and Replenishment output.

 17/6/2025 Today, I will continue to learn more about the structure and logic of the MinMax output file, and further understand the actual role of each field in inventory management and replenishment strategy. At the same time, I will also start to study and analyze the content of the Replenishment file, and try to understand how the entire replenishment script works.

 16/6/2025 Today's main task is to understand the content of the MinMax output file and its business meaning, so that the data can be analyzed and applied more effectively in the future. Read and parse each column of data in the MinMax file one by one, and understand the name, unit and calculation logic of each field.

 13/6/2025 Today i continued to complete the work of integrating holiday factors into the sales forecasting model, and used the new model that included holiday information to predict sales, and then compared and analyzed the prediction results with the original prediction results that did not include holiday factors.

 12/6/2025 Today i are focusing on finding ways to further improve the accuracy of sales forecasts, especially considering incorporating holiday effects into the model to more closely match real sales fluctuations.

 11/6/2025 Today, i will continue to study the parameter adjustment method of the SARIMA model in depth, focusing on finding the parameter combination that best suits the current sales data. By systematically testing different (p, d, q)(P, D, Q, s) configurations and combining the actual sales cycle and seasonal characteristics, i will gradually select several model versions with good performance. After each parameter setting, i will find the model with the smallest MAPE to ensure that the model has higher reliability in practical applications.

 10/6/2025 Today, i successfully completed the construction of the SARIMA model and began to apply it to the prediction of actual sales data. After model training, i generated sales forecasts for the next few days and compared them with the actual data to evaluate the accuracy of the model.

 9/6/2025 Today, I continued to complete the revision of the analysis report. Based on the feedback received in the previous report, I further optimized the presentation of charts, adjusted the analysis logic structure, and ensured that all forecast results were clearer and easier to understand. After completing the revision, I participated in the department meeting again and presented a revised version of the report for internal discussion. After the meeting, I returned to the analysis work and continued to focus on the optimization and application of the SARIMA model.

 Successfully completed the task assigned by Supply Chain Executive to conduct in-depth data analysis and sales forecast for 20 designated products. After completing the task, organize all analysis results and charts, prepare a complete report and present it to relevant personnel.

 5/6/2025 Today, I continued to try to build and optimize the SARIMA model, and tried to improve the prediction accuracy of the model by constantly adjusting the seasonal and non-seasonal parameter combinations. At the same time, I also accepted a new task from Supply Chain Executive: Analyze sales data of 20 different items.

 4/5/2025 After in-depth study of the structure and principles of the SARIMA model, I successfully mastered how to build a complete SARIMA model and began to apply the model to sales forecasting.

 3/5/2025 Continue to study the construction and application of the SARIMA model in depth, and further understand that before using the ACF and PACF graphs to determine the model parameters, seasonal differences must be performed to eliminate the seasonal trend in the time series. After seasonal differences, use the ACF and PACF graphs to determine whether there is still non-seasonal autocorrelation or partial autocorrelation, so as to decide whether a non-seasonal difference is needed.

 2/6/2025 Public holiday