Failure rates in multi-step, manual sequences

· 302 words · 2 minute read

Source Data 🔗

Process 🔗

1 - List the steps 🔗

  1. Navigate to original repository
  2. Right of the Branch mean, click New Pull Request
  3. On the Compare page, click compare across forks
  4. Confirm base-fork and base-branch
  5. Use head fork to select fork, compare branch to select branch
  6. Type a title and description for your pull request
  7. If you do not want to allow upstream edits, unselect Allow edits

2 - Classify the steps and add error rates 🔗

Steps Classification Error rate
1. Navigate to original repository Routine, familiar 0.0004
2. Right of the Branch mean, click New Pull Request Routine, instruction 0.0060
3. On the Compare page, click compare across forks Routine, instruction 0.0060
4. Confirm base-fork and base-branch Non-routine, instruction 0.0300
5. Use head fork to select fork, compare branch to select branch Non-routine, instruction 0.0300
6. Type a title and description for your pull request Routine, familiar 0.0004
7. If you do not want to allow upstream edits, unselect Allow edits Routine with deviation 0.0160

3 - Add success rate 🔗

Steps Classification Error rate (pError) Success Rate (1 - pError)
1. Routine, familiar 0.0004 0.9996
2. Routine, instruction 0.0060 0.9940
3. Routine, instruction 0.0060 0.9940
4. Non-routine, instruction 0.0300 0.9700
5. Non-routine, instruction 0.0300 0.9700
6. Routine, familiar 0.0004 0.9996
7. Routine with deviation 0.0160 0.9840

4 - Calculate probability of success 🔗

Take the product of the success rates to find to likelihood of all steps being completed successfully.

In Excel, Google Sheets,

=ROUND(PRODUCT(pSuccess1:pSuccess7), 2)

For this sequence of steps the likelihood of success is 0.91, or 91%.

Apply for number of likely iterations (e.g. 10) using

=BINOMDIST(10, 10, .91, FALSE)

Example 🔗

Example sheet

Download example spreadsheet,

modeling statistics error-rate