Zwischenbericht geschrieben

[km-nerlich]

For the project:

• take pre trained model and then fine tune it and compare it with the original model
• hugging face provides a lot of models
• auto ml for automatic model training

Midterm Presentation

• quite flexible
• the closer to final presentation the better
• get to the points that are challenging, and where we need help
• we can skip parts where we don't need help (which are straight forward)
• corporate Identity for presentation or own design
  • same for final pres
• tell a story also about the mistakes / what didn't work etc.

General proceeding:

1. discovery
2. trying stuff out and document it
3. come together and check what happendend, what went well, what didn't
4. decide on what to follow → continue with what works well

Paper:

• problem statement
• related work
• research approach
• crisp dm model data mining model
  1. [Business Understanding (Aufgabendefinition)](https://datasolut.com/crisp-dm-standard/#Business-Understanding)
  2. [Data Understanding (Auswahl der relevanten Datenbestände)](https://datasolut.com/crisp-dm-standard/#data-understanding)
  3. [Data Preparation (Datenaufbereitung)](https://datasolut.com/crisp-dm-standard/#data-preparation)
  4. [Modeling (Auswahl und Anwendung von Data Mining Methoden)](https://datasolut.com/crisp-dm-standard/#modeling)
  5. [Evaluation (Bewertung und Interpretation der Ereignisse)](https://datasolut.com/crisp-dm-standard/#evaluation)
  6. [Deployment (Anwendung der Ergebnisse)](https://datasolut.com/crisp-dm-standard/#deployment)
• experiments
• results
• critical reflektion
• always diagnose, why models perform bad
• time schedule
• 25 - 30 Seiten