Monte carlo estimation of completion dates based on past performance using Taskwarrior.
Implements the idea of evidence-based scheduling promoted by Joel Spolsky:
https://www.joelonsoftware.com/2007/10/26/evidence-based-scheduling
Requires Taskwarrior 2.4 or above and Python 3.
Add the user defined attributes (UDAs) totalactivetime, estimatedtime, and velocity to your task configuration:
task config uda.totalactivetime.type duration
task config uda.totalactivetime.label Total active time
task config uda.totalactivetime.values ''
task config uda.estimatedtime.type duration
task config uda.estimatedtime.label Estimated time
task config uda.estimatedtime.values ''
task config uda.velocity.type numeric
task config uda.velocity.label Velocity
task config uda.velocity.values ''
Or equavalently, add them directly to your .taskrc file:
uda.totalactivetime.type=duration
uda.totalactivetime.label=Total active time
uda.totalactivetime.values=''
uda.estimatedtime.type=duration
uda.estimatedtime.label=Estimated time
uda.estimatedtime.values=''
uda.velocity.type=numeric
uda.velocity.label=Velocity
uda.velocity.values=''
(So far, the final attribute is not used, but plans are to cache velocities there, which will allow extending predictions to cover tasks in progress.)
These quantities may be added to the task list report:
task config report.list.labels 'ID,Active,Age,...,Est,Elapsed,Vel,...'
task config report.list.columns 'id,start.age,entry.age,...,estimatedtime,totalactivetime,velocity,...'
Then copy the script taskmc.py into your home directory, or create an alias as follows:
alias taskmc='<directory>/taskmc.py'
This script depends on Python 3.7 or above and on the taskw Python library:
pip install taskw
This script is made much more useful by the taskwarrior-time-tracking-hook python script, available here:
pip install taskwarrior-time-tracking-hook
The pip script for taskwarrior-time-tracking-hook also installs taskw for you.
Set estimates for new tasks with
task add ... estimatedtime:<integer>seconds
Set estimates for already existing tasks with
task <task ID> modify estimatedtime:<integer>seconds
These may be abbreviated task add est:<integer>s and task <task ID> mod est:<integer>s if there are no conflicting UDAs.
Set elapsed times for completed tasks as follows:
task <task ID> modify totalactivetime:<integer>seconds
This may be abbreviated as task <task ID> mod tot:<integer>s.
If taskwarrior-time-tracking-hook is installed, elapsed times may be updated more conveniently by calling task <task ID> start and task <task ID> stop before and after working on a task. These may be called an arbitrary number of times on the same task.
When a task is marked as completed (task <task ID> done), it begins to be used as a basis for estimating the accuracy of the time estimates and predicting the duration of the remaining tasks.
An probabilistic estimate for the time remaining to complete all tasks for which estimated times have been specified may then be estimated by calling the script:
taskmc.py
Because the time remaining is calculated based on the ratio between the estimated times and elapsed time for previously completed tasks, a few tasks with both estimated times and elapsed times must be completed before the output is meaningful.
Delete the UDA configuration:
task config uda.totalactivetime.type
task config uda.totalactivetime.label
task config uda.totalactivetime.values
task config uda.estimatedtime.type
task config uda.estimatedtime.label
task config uda.estimatedtime.values
task config uda.velocity.type
task config uda.velocity.label
task config uda.velocity.values
Remove the Python libraries:
pip uninstall taskwarrior-time-tracking-hook
pip uninstall taskw
Remove any shell aliases:
unalias taskmc
And delete the script:
rm <Path>/taskmc.py