Market QA: Use better lead time features for current sr/so model

Start with GRPT-

Steps(Neeha to fill approach what is tried and when)-

 1. stat based features , understand current features--wip 

2. weighted delay

------------------------------------- 28th April--------

1. add stat features for each anchor date , with Market QA /GRPT data
2. for every anchor date take <=5 batches for the material and market location
   1. first check by material, location if have 5batches in the past from anchor date 
   2. else check by only location, take 5baches by location  
3. all data are present 
4. target :Wednesday ,23rd 2023

— update on 9th May ----

Neeha updated : targeting today (9th May) 

Neeha to generate features data in csv to Vishal ,then Vishal will run the model

Vishal will run model once with new features and once without new features 

 

------ update on 11th May —

memory error

Neeha optimizing code

consider running by batches by anchor date

 

----  update on 17th May – --------

Neeha updated , batches by material ,location and then anchor date 

data is already generated ,Neeha will do another QC and send the files to Vishal  today 

 

– update on 22nd May –

Neeha sending the data to Vishal