feat: implement PERTMethod class with three-point estimation (O, M, P) - #5

[rajesh-puripanda]

Summary

Implements core PERT (Program Evaluation and Review Technique) three-point estimation as described in issue #5.

Changes

Core PERT Logic (networkdiagram/networkdiagram.py):

• Extended Node class with optimistic, most_likely, pessimistic fields, plus computed expected_time, variance, and std_deviation properties
• Created PERTMethod class extending CriticalPathMethod with:
  • add_activity(name, optimistic, most_likely, pessimistic) — automatically computes Expected Time TE = (O + 4M + P) / 6
  • add_activities_relations(activities, O_list, M_list, P_list, predecessors) — bulk add with three-point estimates
  • get_expected_time(), get_variance(), get_std_deviation() per activity
  • get_project_variance(), get_project_std_deviation() — summed along critical path
  • find_critical_path() — uses expected times for critical path determination
  • network_summary() — displays O/M/P/TE/Variance/StdDev for each activity
  • display_pert_distribution() — grouped bar chart comparing O, M, P, and TE estimates via matplotlib
• Bug fix: corrected off-by-one error in CriticalPathMethod.add_activities_relations() — durations[i+1] → durations[i] (first duration was previously skipped)

Exports (__init__.py):

• Added PERTMethod and Node to package exports

Tests (tests/):

• test_pert.py — 20 unit tests covering PERT formulas, variance/stddev calculations, critical path finding, project variance aggregation, and backward compatibility
• test_smoke.py — end-to-end integration tests

Other:

• Added .gitignore for Python cache (__pycache__/) and build artifacts

Usage Example

from networkdiagram import PERTMethod

pert = PERTMethod()
pert.add_activity('O')
pert.add_activity('A', optimistic=2, most_likely=4, pessimistic=6)
pert.add_activity('B', optimistic=3, most_likely=5, pessimistic=7)
pert.add_relation('A', '-')
pert.add_relation('B', 'A')
pert.find_probable_paths()
pert.find_critical_path()
pert.network_summary()

Verification
All 20 unit tests pass:

python -m unittest tests/test_pert.py -v
Existing CriticalPathMethod API is fully backward-compatible.

Closes #5

[kathan-majithia]

Great work on this implementation, Rajesh!

The visual distribution chart is fantastic, and the base Expected Time calculations are solid.

However, I ran this locally against a test network with multiple critical paths and caught a crashing bug in find_critical_path.

• If there are multiple paths, the script appends a list to an existing list of strings, creating a nested array (e.g., ['A', 'B', ['A', 'C']]).
• When network_summary() or get_project_variance() iterates over this, it throws a TypeError: unhashable type: 'list'.
• Could you please update the logic to properly handle multiple critical paths (perhaps by ensuring self.critical_path is always a list of lists, e.g., [['A', 'B'], ['A', 'C']]) and update your loops in network_summary and variance calculations to flatten or target the specific paths?
• Also, make sure you pull the latest main branch so this code integrates cleanly with the new forward_pass() and backward_pass() features. Let me know when you've pushed the fix!