Estandarización:
X ∼ N ( μ , σ 2 ) Z = X − μ σ X \sim N(\mu, \sigma^{2}) \quad \quad Z = \frac{X-\mu}{\sigma} X ∼ N ( μ , σ 2 ) Z = σ X − μ
D 0 ∼ N ( 53 , 5 2 ) comunes D_{0} \sim N(53, 5^{2}) \quad \text{ comunes} D 0 ∼ N ( 53 , 5 2 ) comunes
D 1 ∼ N ( μ 1 , σ 1 2 ) falladas D_{1} \sim N(\mu_{1}, \sigma_{1}^{2}) \quad \text{falladas} D 1 ∼ N ( μ 1 , σ 1 2 ) falladas
C = { la pila es comun } F = { La pila es fallada } C = \{ \text{la pila es comun } \} \quad \quad F= \{ \text{La pila es fallada} \} C = { la pila es comun } F = { La pila es fallada }
0.82688 = P ( D ≥ 47 ) 0.82688 =P(D\geq 47) 0.82688 = P ( D ≥ 47 )
Con D = D= D = Duración de una pila random
P ( D ≥ 47 ) = P ( D ≥ 47 ∣ C ) ⋅ P ( C ) ⏟ =0.7 + P ( D ≥ 47 ∣ F ) ⋅ P ( F ) ⏟ 0.3 P(D\geq 47) = P(D \geq 47\bigm| C) \cdot \underbrace{ P(C) }_{ \text{=0.7} }+ P(D \geq 47 \bigm| F) \cdot \underbrace{ P (F) }_{ 0.3 } P ( D ≥ 47 ) = P ( D ≥ 47 C ) ⋅ =0.7 P ( C ) + P ( D ≥ 47 F ) ⋅ 0.3 P ( F )
0.05746 = P ( D ≥ 60 ) = P ( D ≥ 60 ∣ C ) ⋅ 0.7 + P ( D ≥ 60 ∣ F ) ⋅ 0.3 0.05746 = P(D \geq 60)= P(D \geq 60 \bigm| C) \cdot 0.7+P(D \geq 60\bigm| F) \cdot 0.3 0.05746 = P ( D ≥ 60 ) = P ( D ≥ 60 C ) ⋅ 0.7 + P ( D ≥ 60 F ) ⋅ 0.3
Ahora tendríamos que hacer normalización.
P ( D ≥ 47 ∣ C ) = P ( Z ≥ 47 − 53 5 ) P(D\geq 47\bigm| C) = P\left( Z \geq \frac{47-53}{5} \right) P ( D ≥ 47 C ) = P ( Z ≥ 5 47 − 53 )
= P ( Z ≥ − 1 , 2 ) =P(Z \geq -1,2) = P ( Z ≥ − 1 , 2 )
Como la distribución es simétrica y viendo la tabla:
= P ( Z ≤ 1 , 2 ) = 0.8849 =P(Z \leq 1,2) = 0.8849 = P ( Z ≤ 1 , 2 ) = 0.8849
Resulta ser útil recordar que P ( Z > 1.25 ) = 1 − P ( Z ≤ 1.25 ) P(Z> 1.25)=1 - P(Z\leq1.25) P ( Z > 1.25 ) = 1 − P ( Z ≤ 1.25 )
P ( − 1 ≤ Z ≤ 2 ) = P ( Z ≤ 2 ) − P ( Z ≤ 1 ) P(-1 \leq Z \leq2)=P(Z\leq 2) - P(Z\leq 1) P ( − 1 ≤ Z ≤ 2 ) = P ( Z ≤ 2 ) − P ( Z ≤ 1 )
Aplicando el mismo truco
P ( D ≥ 60 ∣ C ) = 0.08076 P(D\geq 60\bigm| C)= 0.08076 P ( D ≥ 60 C ) = 0.08076
P ( D ≥ 60 ) = P ( D ≥ 60 ∣ C ) P ( C ) + P ( D ≥ 60 ∣ F ) P ( F ) = 0.05746 P(D \ge 60) = P(D \ge 60 \mid C)P(C) + P(D \ge 60 \mid F)P(F) = 0.05746 P ( D ≥ 60 ) = P ( D ≥ 60 ∣ C ) P ( C ) + P ( D ≥ 60 ∣ F ) P ( F ) = 0.05746
Primero, calculamos las probabilidades correspondientes a las pilas comunes estandarizando con Z ∼ N ( 0 , 1 ) Z \sim N(0, 1) Z ∼ N ( 0 , 1 ) :
Primero, calculamos las probabilidades correspondientes a las pilas comunes estandarizando con Z ∼ N ( 0 , 1 ) Z \sim N(0, 1) Z ∼ N ( 0 , 1 ) :
P ( D ≥ 47 ∣ C ) = P ( Z ≥ 47 − 53 5 ) = P ( Z ≥ − 1.2 ) = P ( Z ≤ 1.2 ) = 0.88493 \begin{aligned}
P(D \ge 47 \mid C) &= P\left(Z \ge \frac{47 - 53}{5}\right) \\
&= P(Z \ge -1.2) = P(Z \le 1.2) \\
&= 0.88493
\end{aligned} P ( D ≥ 47 ∣ C ) = P ( Z ≥ 5 47 − 53 ) = P ( Z ≥ − 1.2 ) = P ( Z ≤ 1.2 ) = 0.88493
P ( D ≥ 60 ∣ C ) = P ( Z ≥ 60 − 53 5 ) = P ( Z ≥ 1.4 ) = 1 − P ( Z ≤ 1.4 ) = 1 − 0.91924 = 0.08076 \begin{aligned}
P(D \ge 60 \mid C) &= P\left(Z \ge \frac{60 - 53}{5}\right) \\
&= P(Z \ge 1.4) = 1 - P(Z \le 1.4) \\
&= 1 - 0.91924 = 0.08076
\end{aligned} P ( D ≥ 60 ∣ C ) = P ( Z ≥ 5 60 − 53 ) = P ( Z ≥ 1.4 ) = 1 − P ( Z ≤ 1.4 ) = 1 − 0.91924 = 0.08076
Sustituimos estos valores en nuestras ecuaciones de probabilidad total para aislar la población fallada:
0.7 ⋅ ( 0.88493 ) + 0.3 ⋅ P ( D 1 ≥ 47 ) = 0.82688 ⟹ 0.3 ⋅ P ( D 1 ≥ 47 ) = 0.207429 ⟹ P ( D 1 ≥ 47 ) = 0.69143 \begin{aligned}
0.7 \cdot (0.88493) + 0.3 \cdot P(D_1 \ge 47) &= 0.82688 \\
\implies 0.3 \cdot P(D_1 \ge 47) &= 0.207429 \\
\implies P(D_1 \ge 47) &= 0.69143
\end{aligned} 0.7 ⋅ ( 0.88493 ) + 0.3 ⋅ P ( D 1 ≥ 47 ) ⟹ 0.3 ⋅ P ( D 1 ≥ 47 ) ⟹ P ( D 1 ≥ 47 ) = 0.82688 = 0.207429 = 0.69143
0.7 ⋅ ( 0.08076 ) + 0.3 ⋅ P ( D 1 ≥ 60 ) = 0.05746 ⟹ 0.3 ⋅ P ( D 1 ≥ 60 ) = 0.000928 ⟹ P ( D 1 ≥ 60 ) = 0.003093 \begin{aligned}
0.7 \cdot (0.08076) + 0.3 \cdot P(D_1 \ge 60) &= 0.05746 \\
\implies 0.3 \cdot P(D_1 \ge 60) &= 0.000928 \\
\implies P(D_1 \ge 60) &= 0.003093
\end{aligned} 0.7 ⋅ ( 0.08076 ) + 0.3 ⋅ P ( D 1 ≥ 60 ) ⟹ 0.3 ⋅ P ( D 1 ≥ 60 ) ⟹ P ( D 1 ≥ 60 ) = 0.05746 = 0.000928 = 0.003093
Ahora estandarizamos la variable de las pilas falladas D 1 D_1 D 1 para hallar μ 1 \mu_1 μ 1 y σ 1 \sigma_1 σ 1 :
P ( Z ≥ 47 − μ 1 σ 1 ) = 0.69143 ⟹ P ( Z ≤ μ 1 − 47 σ 1 ) = 0.69143 ⟹ 47 − μ 1 σ 1 = − 0.50 \begin{aligned}
P\left(Z \ge \frac{47 - \mu_1}{\sigma_1}\right) &= 0.69143 \\
\implies P\left(Z \le \frac{\mu_1 - 47}{\sigma_1}\right) &= 0.69143 \\
\implies \frac{47 - \mu_1}{\sigma_1} &= -0.50
\end{aligned} P ( Z ≥ σ 1 47 − μ 1 ) ⟹ P ( Z ≤ σ 1 μ 1 − 47 ) ⟹ σ 1 47 − μ 1 = 0.69143 = 0.69143 = − 0.50
P ( Z ≥ 60 − μ 1 σ 1 ) = 0.003093 ⟹ P ( Z ≤ 60 − μ 1 σ 1 ) = 0.996907 ⟹ 60 − μ 1 σ 1 = 2.75 \begin{aligned}
P\left(Z \ge \frac{60 - \mu_1}{\sigma_1}\right) &= 0.003093 \\
\implies P\left(Z \le \frac{60 - \mu_1}{\sigma_1}\right) &= 0.996907 \\
\implies \frac{60 - \mu_1}{\sigma_1} &= 2.75
\end{aligned} P ( Z ≥ σ 1 60 − μ 1 ) ⟹ P ( Z ≤ σ 1 60 − μ 1 ) ⟹ σ 1 60 − μ 1 = 0.003093 = 0.996907 = 2.75
Tenemos un sistema lineal de dos ecuaciones con dos incógnitas:
47 − μ 1 = − 0.50 σ 1 ⟹ μ 1 = 47 + 0.50 σ 1 60 − μ 1 = 2.75 σ 1 \begin{aligned}
47 - \mu_1 &= -0.50\sigma_1 \implies \mu_1 = 47 + 0.50\sigma_1 \\
60 - \mu_1 &= 2.75\sigma_1
\end{aligned} 47 − μ 1 60 − μ 1 = − 0.50 σ 1 ⟹ μ 1 = 47 + 0.50 σ 1 = 2.75 σ 1
Restando ambas expresiones:
13 = 3.25 σ 1 ⟹ σ 1 = 4 ⟹ σ 1 2 = 16. \begin{aligned}
13 &= 3.25\sigma_1 \\
\implies \sigma_1 &= 4 \implies \sigma_1^2 = 16.
\end{aligned} 13 ⟹ σ 1 = 3.25 σ 1 = 4 ⟹ σ 1 2 = 16.
Sustituyendo el valor de σ 1 \sigma_1 σ 1 : μ 1 = 47 + 0.50 ( 4 ) = 49 \mu_1 = 47 + 0.50(4) = 49 μ 1 = 47 + 0.50 ( 4 ) = 49 .
b) Tenemos que F D ( x ) = F D ∣ C ( x ) P ( C ) + F D ∣ F ( x ) P ( F ) F_D(x) = F_{D|C}(x)P(C) + F_{D|F}(x)P(F) F D ( x ) = F D ∣ C ( x ) P ( C ) + F D ∣ F ( x ) P ( F ) . Derivando respecto de x x x ,
f D ( x ) = f D ∣ C ( x ) P ( C ) + f D ∣ F ( x ) P ( F ) f_D(x) = f_{D|C}(x)P(C) + f_{D|F}(x)P(F) f D ( x ) = f D ∣ C ( x ) P ( C ) + f D ∣ F ( x ) P ( F )
f D ( x ) = 0.7 2 π ⋅ 25 e − ( x − 53 ) 2 2 ⋅ 25 + 0.3 2 π ⋅ 16 e − ( x − 49 ) 2 2 ⋅ 16 \begin{aligned}
f_D(x) &= \frac{0.7}{\sqrt{2\pi \cdot 25}} e^{-\frac{(x-53)^2}{2\cdot 25}} \\
&\quad + \frac{0.3}{\sqrt{2\pi \cdot 16}} e^{-\frac{(x-49)^2}{2\cdot 16}}
\end{aligned} f D ( x ) = 2 π ⋅ 25 0.7 e − 2 ⋅ 25 ( x − 53 ) 2 + 2 π ⋅ 16 0.3 e − 2 ⋅ 16 ( x − 49 ) 2
c) Nos piden hallar la probabilidad condicional P ( F ∣ D > 51 ) P(F \mid D > 51) P ( F ∣ D > 51 ) . Aplicamos la regla de Bayes:
P ( F ∣ D > 51 ) = P ( D > 51 ∣ F ) P ( F ) P ( D > 51 ∣ C ) P ( C ) + P ( D > 51 ∣ F ) P ( F ) P(F \mid D > 51) = \frac{P(D > 51 \mid F)P(F)}{P(D > 51 \mid C)P(C) + P(D > 51 \mid F)P(F)} P ( F ∣ D > 51 ) = P ( D > 51 ∣ C ) P ( C ) + P ( D > 51 ∣ F ) P ( F ) P ( D > 51 ∣ F ) P ( F )
Calculamos las probabilidades de los componentes estandarizando:
P ( D > 51 ∣ F ) = P ( Z > 51 − 49 4 ) = P ( Z > 0.5 ) = 1 − 0.69146 = 0.30854 \begin{aligned}
P(D > 51 \mid F) &= P\left(Z > \frac{51 - 49}{4}\right) \\
&= P(Z > 0.5) = 1 - 0.69146 \\
&= 0.30854
\end{aligned} P ( D > 51 ∣ F ) = P ( Z > 4 51 − 49 ) = P ( Z > 0.5 ) = 1 − 0.69146 = 0.30854
P ( D > 51 ∣ C ) = P ( Z > 51 − 53 5 ) = P ( Z > − 0.4 ) = P ( Z < 0.4 ) = 0.65542 \begin{aligned}
P(D > 51 \mid C) &= P\left(Z > \frac{51 - 53}{5}\right) \\
&= P(Z > -0.4) = P(Z < 0.4) \\
&= 0.65542
\end{aligned} P ( D > 51 ∣ C ) = P ( Z > 5 51 − 53 ) = P ( Z > − 0.4 ) = P ( Z < 0.4 ) = 0.65542
Reemplazamos en la fórmula de Bayes:
P ( F ∣ D > 51 ) = 0.30854 ⋅ 0.3 0.65542 ⋅ 0.7 + 0.30854 ⋅ 0.3 = 0.092562 0.458794 + 0.092562 = 0.092562 0.551356 ≈ 0.1679 \begin{aligned}
P(F \mid D > 51) &= \frac{0.30854 \cdot 0.3}{0.65542 \cdot 0.7 + 0.30854 \cdot 0.3} \\
&= \frac{0.092562}{0.458794 + 0.092562} \\
&= \frac{0.092562}{0.551356} \\
&\approx 0.1679
\end{aligned} P ( F ∣ D > 51 ) = 0.65542 ⋅ 0.7 + 0.30854 ⋅ 0.3 0.30854 ⋅ 0.3 = 0.458794 + 0.092562 0.092562 = 0.551356 0.092562 ≈ 0.1679