Just a first try to connect with ML.NET. For read only and evaluation purposes.
Dataset for beginers in ML provided by Kaggle. Attempt to predict who will survive. Only age + sex + ticket price was used as input data, other data as port, ticket number was discarded as useless. Dataset prepare strategies:
- Discard missed age data
- Fill missed age data by mean
Algo | Data Preparation | Result |
---|---|---|
OneVersusAll Learner + Averaged perceptron | Missed data discarded | 65.55% |
OneVersusAll Learner + Averaged perceptron | Missed data filled by mean | 64.832% |
Pairwise coupler + LinearSvm | Missed data discarded | 63.397% |
Microsoft Model Builder (FastTree) | Missed data discarded | 50.956% |
Attempt to forecast bitcoin prices. Long story short: mean error 12%, but good prediction on big time intervals.
Anomaly detection algorithm in heart rate data. Strategies/hypothesis:
- Anomaly detection based on cycade rythm
- Anomaly detection based on physical activity
Long story short: all strategies/hypothesis are good/confirmed