So you’ve done it, you’ve got a nice time series with helpful features in a
pandasDataFrame. Maybe you’ve used
pd.bfill() to fill in empty time steps using the previous or next value and perform analysis or feature extraction on your full series.
What do you do?
We faced this problem today at Mansa, where we were saving hundreds (if not thousands) of unnecessary rows after our pre-processing pipeline was completed.
Since we deal with financial data, we want to be able to tell the balance for an account at any point in time and to calculate statistics over number…
MSc. in AI @ UVA | Previously, LD15 @ EF and 1st Employee + ML Engineer @ Mansa