SAT-based Staff Scheduling

This project implements a constraint-based staff scheduling system using a SAT solver.
It generates and transforms work schedules for 20 people across multiple weeks, under strict rules of work distribution and rest.

---

Features

• SAT formulation of the scheduling problem:
  • 20 people assigned to stations over 14 days per week.
  • Different numbers of daytime and nighttime stations.
  • Constraints on workload, rest, and distribution.
• Exportable schedule:
  • planning_table.csv: station-based assignments.
  • planning_par_personne.csv: person-based assignments with statistics.
• Transformations & Statistics:
  • Number of days, nights, and Sundays worked per person.
  • Easy-to-read pivot table (per person × per day).

---

Constraints

The schedule must satisfy:

1. Each station is occupied by exactly one person.
2. A person cannot be assigned to more than one station per day/night.
3. Maximum 4 worked days among days 1–6.
4. Minimum 2 worked days among days 1–6.
5. Maximum 3 assignments between Sunday (day 7) and nights (days 8–14).
6. Minimum 1 assignment between Sunday (day 7) and nights (days 8–14).
7. Mandatory rest: if a person works a night, they cannot work the following day.

---

Project Structure

.
├── planning\_sat.py              # SAT encoding and solving
├── planning\_transformer.py      # Transform station-based to person-based planning
├── planning\_table.csv           # Example output (station-based)
├── planning\_par\_personne.csv   # Example output (person-based with stats)
└── README.md                     # Project documentation

---

Dependencies

• PySAT – SAT encoding and solving
• pandas – data manipulation
• itertools – combinatorial generation (standard library)
• re, string, csv – standard Python libraries

---

Usage

1. Generate a schedule with SAT solver

python planning_sat.py

This creates the file:

• planning_table.csv → schedule organized by station.

2. Transform into person-based planning

python planning_transformer.py

This creates:

• planning_par_personne.csv → schedule organized by person, with workload stats.

---

Example Output

Station-based (planning_table.csv)

station	S1_day1	S1_night1	...
CD1	A		...
CD2	B		...
...	...	...	...

Person-based (planning_par_personne.csv)

person	S1 - Monday - day	S1 - Monday - night	...	nb_jours	nb_dimanche_jour	nb_nuits
A	CD1		...	3	1	2
B	CD2		...	4	0	1

---

TODO

• Add option for a free week
• Separate Sunday and night constraints
• Refactor into OOP (object-oriented design)
• GUI interface for easier usage
• Better error handling and file management