1
$\begingroup$

I perform $n$ simulations of operations planning. I can plot a gantt chart for each simulation, but I am interested in visualising the variability of the whole simulation set : I want a Gantt visualisation which includes information about :

  • the distribution of starting dates $d_s$
  • the distribution of durations $\Delta$ (or ending dates $d_e = d_s + \Delta$)

I thought of a plot derived from Gantt Plot where each process is represented by a box with information showing mean and specific quantiles for starting dates and ending dates (inspired by boxplots).

sketch

Do you have other ideas / recommandations on how to visualize the variability of a task's properties in a Gantt chart ?

Additionnally, do you have any idea of how to implement this in Python ? Ideally as an interactive visualisation ? I thought of creating a new plotly class derived from their Boxplot.

Thanks :)

EDIT : here is a data sample for starting dates (in real life there are thousands of rows -simulations- and hundreds of columns -processes):

key_op     1     2_FS     3_SF     4_FF     5_SS     6_FS     7_SF     8_FF     9_SS     10_FS     11_SF     12_FF     13_SS
    0     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    1     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    2     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    3     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    4     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    5     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    6     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    7     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    8     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00
    9     5/1/18 0:00     5/1/18 3:00     4/1/18 21:00     5/1/18 0:00     5/1/18 0:00     6/1/18 3:00     5/1/18 21:00     6/1/18 0:00     6/1/18 0:00     4/1/18 3:00     3/1/18 21:00     4/1/18 0:00     4/1/18 0:00

ending dates:

key_op   1   2_FS   3_SF   4_FF   5_SS   6_FS   7_SF   8_FF   9_SS   10_FS   11_SF   12_FF   13_SS
    0   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    1   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    2   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    3   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    4   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    5   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    6   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    7   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    8   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00
    9   5/1/18 3:00   5/1/18 6:00   5/1/18 0:00   5/1/18 3:00   5/1/18 3:00   6/1/18 6:00   6/1/18 0:00   6/1/18 3:00   6/1/18 3:00   4/1/18 6:00   4/1/18 0:00   4/1/18 3:00   4/1/18 3:00

and links between operations based on :

op_id  previous_op
1  
2  1FS
3  1SF
4  1FF
5  1SS
6  1FS+1 day
7  1SF+1 day
8  1FF+1 day
9  1SS+1 day
10  1FS-1 day
11  1SF-1 day
12  1FF-1 day
13  1SS-1 day
$\endgroup$
8
  • $\begingroup$ Do you really only have 2 values (start date & duration)? If so, a gantt chart seems like overkill. Gantt charts are typically sensible when there are multiple processes to visualize / keep track of, and their starting & ending points are linked between processes. $\endgroup$ Apr 26, 2021 at 13:50
  • $\begingroup$ I have several processes in a planning, which are represented by their respective starting date, duration, ending date, and by their links to other processes. However visualizing the links between operations is a bonus, what I am mostly interested in is the variability on starting dates, duration and ending dates for each operation in a Gantt style visualisation $\endgroup$
    – Clément F
    Apr 26, 2021 at 15:38
  • $\begingroup$ Can you provide a small, example dataset for people to work with? It really sounds to me like the right visualization is a scatterplot of duration vs start date. $\endgroup$ Apr 26, 2021 at 15:42
  • $\begingroup$ Thanks, I just added sample datasets to my question. I am not sure that scatterplots would bea readable if there is a big amount of simulations. Besides, I am looking for a visualisation which is readable by non-statisticians. $\endgroup$
    – Clément F
    Apr 26, 2021 at 16:09
  • $\begingroup$ I'm not sure if I get it. Does a simulation entail many sub-processes, with each row an independent simulation & each column is a sub-process? Should there be hours? (These all seem to be on the same day.) $\endgroup$ Apr 26, 2021 at 17:06

0

Your Answer

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