DATA QUALITY IMPROVEMENT FOR ANGKOT ACTIVITY ANALYSIS AND THE CREATION OF ANGKOT BEHAVIOR DETECTION ALGORITHM
In previous studies geolocation data has been collected through two sources, namely the angkot Android application and the WiFi Module ESP8266 as a GPS tracker. The collected data is stored in a database that is implemented using Mongo DB. The stored data was then observed and found 15 types of &...
Saved in:
| Main Author: | |
|---|---|
| Format: | Theses |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/70695 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | In previous studies geolocation data has been collected through two sources,
namely the angkot Android application and the WiFi Module ESP8266 as a GPS
tracker. The collected data is stored in a database that is implemented using Mongo
DB. The stored data was then observed and found 15 types of 'dirty data' based on
Lin Li's dirty data taxonomy. With the discovery of 15 types of 'dirty data' on angkot
data, cleaning actions need to be taken in order to obtain accurate analysis results.
Before cleaning the data, measurement of the quality of the data is done using
Redman's theory. After measuring the quality of angkot data, the process of
improving data quality can be done by data cleaning. The data cleaning process is
adjusted to the purpose of data processing, which is to analyze the activity of angkot
based on the data that is owned and also adjusted to the condition of the existing
data. The data cleaning method applied to the angkot data successfully cleared
86.6% of the previous types of ‘dirty data’.
After the angkot data goes through the data cleaning process, the final results of
the angkot data from the data cleaning process will immediately follow the data
pre-processing stages, the output of this process is the angkot data that is ready to
be displayed on the web application as the final result of the data visualization
process. In a web application, angkot activity on the data will be seen, also angkot
travel patterns can be observed. From here, the normal and abnormal angkot
behavior can be defined to help create angkot behavior detection algorithms. After
the angkot behavior detection algorithm has been developed, the experiment is
conducted by comparing the same abnormal time between the results of the angkot
behavior detection algorithm with the results of exploration on a web application.
The accuracy of the angkot behavior detection algorithm is 63%. |
|---|