My question is how to assign variables within a loop in KQL magic command in Jupyter lab. I refer to Microsoft's document on this subject and will base my question on the code given here:
https://docs.microsoft.com/en-us/azure/data-explorer/kqlmagic
1. First query below
%%kql
StormEvents
| summarize max(DamageProperty) by State
| order by max_DamageProperty desc
| limit 10
2. Second: Convert the resultant query to a dataframe and assign a variable to 'statefilter'
df = _kql_raw_result_.to_dataframe()
statefilter =df.loc[0].State
statefilter
3. This is where I would like to modify the above query and let statefilter have multiple variables (i.e. consist of different states):
df = _kql_raw_result_.to_dataframe()
statefilter =df.loc[0:3].State
statefilter
4. And finally I would like to run my kql query within a for loop for each of the variables within statefilter. This below syntax may not be correct but it can give an example for what I am looking for:
dfs = [] # an empty list to store dataframes
for state in statefilters:
%%kql
let _state = state;
StormEvents
| where State in (_state)
| do some operations here for that specific state
df = _kql_raw_result_.to_dataframe()
dfs.append(df) # store the df specific to state in the list
The reason why I am not querying all the desired states within the KQL query is to prevent resulting in really large query outcomes being assigned to dataframes. This is not for this sample StormEvents
table which has a reasonable size but for my research data which consists of many sites and is really big. Therefore I would like to be able to run a KQL query/analysis for each site within a for loop and assign each site's query results to a dataframe. Please let me know if this is possible or perhaps there may other logical ways to do this within KQL...
question from:
https://stackoverflow.com/questions/66056001/parametrize-and-loop-kql-queries-in-jupyterlab 与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…