Hi Guys,
Need your help in choosing a algorithm which helps in predicting future 1 month data of cpu and memory for esx_cluster whose value is dependent on host and vm metrics ( cpu and memory ) I have tried Kalman filter and LPP5 algorithm by feeding one year data and got some difference between actual value and predicted value PFB query I have used
my_query(with timechart at end)| predict "esx_cluster_name" as prediction algorithm=LLP5 holdback=2 future_timespan=32 upper95=upper95 lower95=lower95 period=30| `forecastviz(32, 2, "esx_cluster_name", 95)`
By using above query I got good R square statics around 0.95 and RMSE around 0.7
Please help me with either choosing a different algorithm which takes the dependencies mentioned above or by suggesting me how to improve my current query to get better prediction's so that difference between actual value and predicted value reduces
Thanks in advance,
K.Rohit
Happy Splunking
