I have checked the studio…thats the cool app you guys have built…thanks for that.
For anomaly detection,It suggested me to use aggregation of 30 min.i have few queries on this.
1)If i ll use aggregation…i m afraid that some of the actual anomaly might be skipped i,e there will be false negative
2)somwhere i read it is just learning from 500 records and accordingly suggesting the aggregation.Is my understanding correct?
3)I m aiming for both prediction as well as anomaly …I m getting very poor results from the swarmed parameter.So i wanted to tune different parameters in model params.However I m not sure which parameters to be changed and how much…is there any reading material available for tuning params for both prediction as well as anomaly?