Hology 7.0 · 1st Place · Data Analyst · Data & Algorithms
Hology: Marketplace Insights
Marketplace analysis and a seller dashboard for our Hology-winning entry.
Background
For Hology 7.0’s Data Mining competition, our team SySendiri explored product discovery in the TokoGenZ marketplace. Poorly organized listings made relevant products harder to find, while sellers lacked a clear view of product categories and performance. The source dataset contained 933,985 product records; clustering used a sample.
Key Decisions
Check What the Data Can Support
The analysis exposed missing location, rating and sales values, and location entries that were actually shop names. The presentation called out these issues and identified stricter validation as a next step.
Give Sellers a Category-Level View
The dashboard combined product counts, average prices, sales and ratings with a category filter. This let a viewer explore a specific category instead of relying only on marketplace-wide totals.
Explain the Models Through Their Outputs
We presented examples from 11 product groups and recommendations based on product-name similarity. The recommendation examples showed related fabrics, dictionaries and laundry products, making the output easier to assess alongside the method.
Results
Our team won 1st place in Hology 7.0 Data Mining. The presentation documents a seller dashboard, sample clustering results and similar-product recommendations. The award recognizes the competition work; the presentation does not establish changes in marketplace sales or shopper behavior.
Scope & Limitations: The 11-cluster configuration was a prototype choice, not a proven optimum: the later analysis notes that more clusters could improve separation. Recommendations used product-name similarity, not purchase history or personalized preferences. Further work included stricter data validation, more specific product groups and richer seller analytics.
Contribution
I focused on data analysis, the seller dashboard and the presentation. My teammate Mohamad Maulana focused on the data-science work. Together, we connected the analysis to three outputs: product groups, seller insights and similar-product recommendations.