LaserChallenge - Team AAA

• 

  Team AAA/YYYYYYYYYYYYYYYYYYYYYYYYYY

Context (German)

Bei diesem Hackathon sollen Ideen mit Artifical-Intelligence-Bezug in 24 Stunden umgesetzt werden. Bildet Teams von 2-5 Personen und baut rund um die Uhr an eurem Prototypen. Die fertigen Projekte werden dann am Schluss vor einer Jury vorgestellt und kämpfen um das Preisgeld im Wert von 10.000 €.

What it does

Our tool that chooses the best parameter sets from a list of possible process parameters with a ~98% success rate!

How We built it

We used

• Genetic Algorithms
• AdaBoost
• GradientBoosting
• RandomForests
• XGBoost
• CatBoost
• t-SNE

Challenges I ran into

The data we analysed was very noisy in the beginning.

Accomplishments that I'm proud of

We reduced our set of prediction features from 100 to 15. With this set of features, we achieved a specificity of ~98%.

What I learned

When training our models, the most important factor was how good we knew our data.

What's next for LaserChallenge

Create even easier explanations for using our models with real world data.

Built With

• google-cloud
• jupyter-notebook
• python
• sklearn