I am new to AI/ML and have attended lots of audit only training on coursera related to AI/ML domain and would like to implement those concepts on real data of product I am working on so that I learn along with doing something fruitful. I am struggling with Data preprocessing and need some suggestion. I have data of few PM counters collected every 15 mins for whole day, and I have lots of this kind of data collected per day and we often see system restarting because of board temperature exceed for long duration or CPU/MEM stats crossing thresold over the period so would like to have a system which can predict system restart.
|KPI | 0:15| 0:30| ..... |24:00|
|:----|:--------|:-----|....... |:----|
|mem | 34 |23 |...... |:10 |
|cpu | 90 |56 |..... | 0 |
|Btemp| 76 | 12 |..... |63 |
|xyz | | |..... | |
.....
So I have csv file with first row as column name and first column values are key/label for data which proceed in next 24x4 columns for data collected for that label every 15 mins. How Can I prepare data out of this csv file using tensorflow etc. I would like to infer/predict malfunctioning based on few KPIs ( not all).