Dash app to compute order prices for brass rods (a copper : zinc alloy).
Client pricing depends on the weight, rod length, the ratio of copper : zinc and their respective material prices.
The calculation for order prices is performed in a directed acyclic graph. Each of the order parameters represents a node on the graph, and the computation is split into single assignment code. The benefit of this approach is that we obtain the order price as well as its derivative w.r.t. each input parameter. This allows us to do a sensitivity analysis to see the effect of price shocks of the raw materials on the final order price without repeating the original calculation. The price shock effects are evaluated through the partial derivatives computed in the node graph. This approximation becomes exact when the price model is linear (as it is here).
In short, a single graph traversal yields both the final price output + sensitivity.
Ensure Docker and Docker Compose are installed on the system. Installing Docker Desktop is recommended to achieve this.
Create and run the production Docker container with:
docker-compose up aad-pricing --build
Open an internet browser and navigate to http://127.0.0.1:7071/ to see the UI on the local host.
Run the tests through:
docker-compose up -d aad-pricing-test-local
This action creates a Docker container only for running tests. The image is created from prod + dev dependencies and only changes whenever the dependencies are updated. The test code is mounted into the container, and test results are exported through volume mounts. See the docker-compose.yml for details.
Test results are exported to an Allure xml file on host:
/test-reports/test-results.xml
From /:
poetry build --format wheel; pip install dist/*; poetry run python applications/server.py
Before the price calculation, we need some parameters. Below data field can be changed if desired:
- Copper fraction slider
- Labour factor table
- Copper price
- Zinc prices
- Order parameters (weight, rod length, zinc quality)
- Sensitivity analysis price shock range
Press the Calculate prices button to compute the total price and sensitivity calculation. The sensitivity calculation uses algorithmic differentiation under the hood in the pricing/ComputationalNode and pricing/DirectedAcyclicGraph classes.
Ensure dependencies are installed and written to the poetry lock file by running poetry install; poetry lock
Run unit tests from the root folder through poetry run pytest
Favicon created by Surang: Metal icons created by surang - Flaticon