Commit c92c62e1 authored by uouzl's avatar uouzl
Browse files

Update README.md

parent 91241b19
......@@ -82,9 +82,27 @@ Second Model: 3-dim. Clustering with pre height before emptying, mean emptying i
**Regression**
XGB Regression with grid search to find the best parameters.
XGB Regressor for prediction of heigth for a container at a specific time. Parameters were optimized by a Grid Search.
For Visualization shap.TreeExplainer is used.
Results:
- MSE: 170.76
- Mean Error: 9.5
Results of the relevant predictions (prediction or original value > 100):
![relevantPredictions](https://git.scc.kit.edu/ufesk/bda-analytics-challenge-template/-/raw/master/notebooks/pictures/relevantPredictions.PNG)
We used Shapley Values to extract the Feature Importance:
![featureImportance](https://git.scc.kit.edu/ufesk/bda-analytics-challenge-template/-/raw/master/notebooks/pictures/featureImportance.png)
This plot also shows the impact of a feature on the model prediction:
- Each dot represents one data instance
- Red dot -> high feature value
- Blue dot -> low feature value
- X-Axis position indicates the impact on the output model
![image-20210705112027118](https://git.scc.kit.edu/ufesk/bda-analytics-challenge-template/-/raw/master/notebooks/pictures/image-202107159.PNG)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment