Acer Incorporated, Senior Data Specialist , Jul 2019 ~ NOW

 GFK.NPD.ICT Data Integration And Provides Insight

Use sales data to design a dashboard, and can judge the business sales's KPI in each region. Find out Acer's Market Share averages is in 4th-5th.

 Google Analytics For Consumer Group Analysis

Understand the browsing groups of demographic and interest, find that 35–54 Male visitors most engaged to our Acer website. And also analyze traffic sources and medium, calculate the session, bounce rateand pageviews of the webpage, and combine GSC data and SimilarWeb data to diagnose the webpage.

 UEIP User Experience Improvement Program

Find insights from user's usage information data. Also cooperate with PM to find out only 5% of gaming model have use the extended storage, so we can replace the space with batteries to increase consumers' willingness to buy. And using hierarchical clustering to distinguish Acer's products, and then carry out accurate promotion.

 UEIP Data Use ML Model To Evaluate The Probability Of Product Inactive

Use RandomForest to predict the probability of not using the device for three consecutive months. And the average accuracy is about 78%. Finally, combined with the CRM sending email system to judge whether the list of recommended devices matches the probability of clicking on the email, and can also lock the target to save the cost of sending emails.

 Design Inventory Dashboard And Avg. Days Of Device Inventory

Provide inventory dashboard for each region. To prevent a large number of stocks but shortages in other regions.

ASUSTek Computer Inc., Senior Data Engineer, Nov 2015 ~ Aug 2018

 Activation Heat Map (query country)

Want to know the sales situation in each region, and presented on the Heat Map. As a reference for the effectiveness of advertising. Among them, the latitude and longitude information is converted into national counties and cities.

 Activation Prediction And Sales Volume Prediction And Variable Design

Design and analyze the variables for predicting Activation volume, and conversion the related variables for prediction. It is like making Weather. Survival rate. Activation average change. CPU variable. 

Reduce 7%-50% stock volume per mobile model for different countries.

 User Behavior And Gender Predictions
Collect users various types of data for analysis, and hope to get the difference between male and female users from these information hidden in the data. And then accurately push the appropriate advertising information.

Ultimately, the accuracy is between 70%-80% for different countries.