Offering a non-destructive and non-invasive analytical cannabis testing device. The company utilizes spectrography to overcome challenges posed by the heterogenous cannabis flower.
Estimation of cannabinoid percent by analysis of (cannabis) spectrographic data.
Small data – ~1k flowers, 51 features.
Each flower is measured several times in a spectrographic device
R programming language, MySQL, and Shiny for a web-based application development.
Business Impact and Insights
Two applications were implemented in favor of this project:
- Estimation Application: based on an optimized machine learning algorithm, the application reads spectrograph data and predicts corresponding cannabinoid percentage.
- Model Generation Application: an extensive application that allows its end-users to examine the flower database (filter by date, supplier, cannabis variety, etc.) and generate a customized machine learning model according to specified data selection.