Related Posts
Do we really have lay offs happening in zs ?
Who billed the most today?
Openings at Accenture for various levels
Additional Posts in DataScientists Network
Multiple L5 openings for data analytics in Amazon Bangalore and Hyderabad. Preferred tier 1 college candidates with 5+ years experience in Data Analytics, Business Intelligence Or Data science. Mba tier1/2 candidates with 2+ years experience can also apply.
Do share your profiles at tanmay56789@gmail.com.
Subject: Profile - Experience - Current Company- College
11 likes for dm please
New to Fishbowl?
unlock all discussions on Fishbowl.
Since you mention time series regression problem, curious to know what the missing values are.. by definition, time series is taken at definite interval(most of the cases), and if you have missing data there, it might mean that the observations were not recorded? What kind of data are you working with?
yes target column values were 95% missing, rest of the column values were all present..
How much data do you have in 5%.
Also is that all recent data?
no, that’s not recent even.. totally random 🤯
95% missing data , is it just for dependent variable ? If thats the case change dataset ? U can add categorical feature for the 5% and check if its working .. but ...
If I were you I would have did following things .
1) if it's time series data identify trends for available data and scrap the data from the internet ( if suitable data available ) and create custom dataset show it to to mgmt and ask them for updates , create a model and show them how the prediction can go wrongs ..
2) shows model scores or enjoy looking outside the 🪟
Are you doing AR model or VAR.
Timeseries data will have major implications on missing data... and you need atleast 4 years or 4 complete cycle of data for some proper analysis..
If u are doing smote or any other technique it might fill data.. but loose the essence of it.. i would say try to understand the business or ask an sme to fill the data for you.