I need to generalize a single sequence from N sequences entailing farming tasks/operations and ultimately plotting it on Gantt chart.

There are a total let's assume N sequences = n (total fields) * t (number of years)

Which ML/deep learning algorithm can learn from N sequences and create a single sequence that best represents overall data.

Each field has tasks/operations occurring in a sequence eg. A,B,C and respectively requiring t days (duration variable).

# Sequence for field 1 in year 2019
field1 = pd.DataFrame({"task name": [A,B,C], "start": [pd.Timestamp("2019-04-09 12:00:00"),pd.Timestamp("2019-05-01 12:00:00"), pd.Timestamp("2019-06-03 12:00:00")], "end": [pd.Timestamp("2019-04-14 23:37:56 "),pd.Timestamp("2019-05-06 00:06:25"),pd.Timestamp("2019-06-04 00:08:13")], "duration": [6 days, 6 days,2 days,]})

Similarly there are other N sequences for each n fields.

Tricky part of the problem: 1. Data is not homogenous, since fields belongs to different farmers and differ in farm size. Minority of tasks are varying across fields

My final objective is to plot a Gantt chart of a single sequence that best represents/indicator of sequence of tasks and its respective duration for a crop used by a farmer.

  • $\begingroup$ The description of your data is good, but I cannot completely understand your objective. "best represents" is a very broad term. If you have the same number of tasks for each field in the same order, you might try to simply average they respective start/end time of year. $\endgroup$
    – Valentas
    Jul 2, 2021 at 7:05


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.