This repository manages the files used in the empirical study of the paper: Building and Validating DNN Models for Forecasting the Quality of Cloud Servicest
The following section presents the results obtained for the 16 variables used to develop the experiment, as well as additional information related to them.
- Dataset
- Jupiter Notebook
- BI-GRU Accuracy Metrics
- LSTM Accuracy Metrics
- BI-GRU model predictions
- LSTM model predictions
- ARIMA model predictions
- AUTO ARIMA model predictions
Representation of the datasets with the observations of 16 QoS metrics extracted from the cloud service SAlert monitoring.
The notebook is available on the link, and the directories required for execution are detailed on the readme.txt file
The statistics used are available on the link Model statistics
BI-GRU model performance evaluation using accuracy metrics RMSE, MAE, MAPE
LSTM model performance evaluation using accuracy metrics RMSE, MAE, MAPE
BI-GRU Model fitting with prequential cross-validation
LSTM Model fitting with prequential cross-validation
ARIMA Model fitting with prequential cross-validation
ARIMA Model auto fitting with prequential cross-validation
Variable | Iteration 1 | Iteration 2 | Iteration 3 |
---|---|---|---|
Free Memory | |||
Used Memory | |||
Free Disk | |||
Used Disk | |||
Disk read/s |