PERANCANGAN SISTEM PENENTUAN FREE MAP ACCESS UNTUK FREEMIUM USER PADA PT MAPID DENGAN METODE DATA MINING
PT Multi Areal Planing Indonesia (MAPID) is a technology company that offer service cloud computing platform that develops Geographic Information System tools to collect, manage, visualize, and analyze geospatial data which has been operating since 2018. Business to Customer (B2C) is one of the s...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/65526 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | PT Multi Areal Planing Indonesia (MAPID) is a technology company that offer service
cloud computing platform that develops Geographic Information System tools to
collect, manage, visualize, and analyze geospatial data which has been operating
since 2018. Business to Customer (B2C) is one of the sectors served by PT MAPID. In
running it, PT MAPID applies a freemium business model by providing free map
access to users who successfully register on the PT MAPID platform. From January
2020 to September 2021, the conversion rate from freemium users to subscriber
licenses is very low, at 0.94%. The main reason is that PT MAPID does not yet have a
decision support system for provide free map access because it has not utilized user
data. This research was conducted to build a predictive model for user purchase
decisions using user data. Furthermore, an application prototype will be built that can
be used to run the model.
The methodology in this study refers to the Cross-Industry Standard Process for Data
Mining (CRISP-DM). Alternative algorithms used in the model development process
are Support Vector Machine (SVM), Random Forest, and Gradient Boosting. To
handle cases of imbalanced data, SMOTE and ADASYN oversampling techniques are
applied. The modeling and prototyping process is carried out using Python as a
programming language.
Based on the research that has been done, the best model is obtained with the Gradient
Boosting algorithm, with the ADASYN oversampling dataset, with an accuracy value
of 81%, the precision of 83%, recall of 62%, and f1 of 71%. The model is then
implemented on an application prototype based on a Graphical User Interface (GUI).
The results of this study are predicted to increase the number of users who buy PT
MAPID product licenses by 83%.
|
|---|