Documentos de Académico
Documentos de Profesional
Documentos de Cultura
One must not expect the same error rate for all types of problems, so
its important to classify degree of difficulty. Am I dealing with a GE1
(low risk) project or a GE3 (high risk) project? GE3 projects would
have high costs and require a higher level of geotechnical rigour to
complete successfully compared to GE1 projects.
Figure 3 shows the statistics for the hard puzzles I solved. My best to
worst time varied from 15 to 88 minutes, respectively, with an
average of 42 minutes for 66 hard puzzles from the Daily Mail. This
outcome provides many more insights compared to the easy Sudoku
puzzles. As illustrated in Figure 3, the best-fit PDF is a log normal
distribution compared to a gamma distribution for the easy puzzles.
But for engineering purposes, they were approximately similar as
seen in Figure 2. The normal PDF was the 9th-ranked best fit in the
goodness-of-fit test.
This clearly shows the normal distribution is not a good predictor for
hard Sudoku puzzles. As the normal PDF is symmetrical, then the
longer times taken to complete the puzzle provides data to compute
corresponding worse times (incorrectly) to balance the normality
assumption and maintain symmetry. The best-fit log normal PDF
shows zero percent chance of a 5-minute solution time, as compared
with the normal distribution prediction. At the 25 percent level, the
PDFs provide an approximately similar predicted time.
The inaccuracy results for the easy, medium, and hard puzzles were 9
percent (1/11), 18 percent (1/5), and 45 percent, respectively. The
COV for easy, medium, and hard were 25 percent, 32 percent, and 54
percent, respectively. Comparing these values with Table 1 shows
similarities where less data produces greater variability. More
challenging puzzles, like more difficult projects, have a larger time
variance, and thus are more likely to be inaccurate and take longer
time to complete.
Estimates of times can be skewed. Below-average time events may
take nominally less time, while time estimates that are above the
average, would not be normally distributed (nominally longer), but
will have a significant error. Hence, when we get it wrong, it will be
wrong in a big way.
There is a common assumption that a normal distribution applies to
all geotechnical data. For example, in theProceedings, 18th ICSMGE,
Look and Lacey (2013) illustrate the erroneous characteristic design
strength that may be produced with a normal PDF for high COVs. This
Sudoku distribution also shows how assuming a normal distribution
should only be applied if the COV is below 25 percent. As the COV
increases, so does the need to apply a best-fit distribution function
rather than assuming a normal distribution applies.
Incomplete Data and Time Constraints
The difference between a hard vs. easy Sudoku puzzle is less data to
start the evaluation process. The game is based on the concept of
Sudokus Lessons
Conflicts between quality, cost, and time often lead to compromises,
and hence increased project risk. If time becomes a constraint in
solving a Sudoku puzzle, then some findings from solving these
puzzles can offer insights about how we deliver projects. They
include:
Thus, if we deliver projects below time and cost, the outcome will be
nominally less, but when we do not meet the project schedule, the
overrun can be significant.
Sudoku may be entertaining, but the game can also teach us many
lessons. For example, incomplete data increases the likely error. The
same is true in geotechnical engineering. In Sudoku there is a unique
solution, while geotechnical engineering is far more complicated, with
several plausible solutions. And Sudokus most important final lesson
is take a break! Constantly pushing to get a solution faster
increases both stress and errors.
TRADUCIDO :
Lognormal (fcil de entender por los ingenieros segundo-mejor ajuste PDF a estos datos)
Normal (ms comnmente utilizados por los ingenieros noveno-mejor ajuste PDF a estos datos)
(Tabla 1 Los rangos de variacin de la propiedad del suelo (Phoon y Kulhawy, 2008))
Lecciones de Sudoku
Los conflictos entre la calidad, el costo y el tiempo a menudo
conducen a compromisos, y por lo tanto, el aumento de los riesgos
del proyecto. Si el tiempo se convierte en una restriccin en la
solucin de un rompecabezas Sudoku, a continuacin, algunos de los
hallazgos de la resolucin de estos rompecabezas pueden ofrecer
ideas sobre cmo entregamos proyectos. Ellos incluyen: