Residuals
If we look at the plot that shows the dependence of flavanoids on the number of phenols, it will be obvious that the use of linear regression, in this case, was not entirely correct. Moreover, how do we interpret how good our prediction is?
Some points will lie on our constructed line, and some will lie away from it. We can measure the distance between a point and a line along the y-axis. This distance is called the residual or error. The remainder is the difference between the observed value of the target and the predicted value. The closer the residual is to 0, the better our model performs. Let's calculate the residuals and present them as a chart.
12345678residuals = Y_test - y_test_predicted # Visualize the data ax = plt.gca() ax.set_xlabel('total_phenols') ax.set_ylabel('residuals') plt.scatter(X_test, residuals) plt.show()
Output:
Our residuals formed three almost straight lines. This distribution is a sign that the model is not working. Ideally, the remains should be arranged symmetrically and randomly around the horizontal axis. Still, if the residual graph shows some pattern (linear or non-linear), it means that our model is not the best.
Swipe to start coding
Try to find residuals to our previous challenge:
- [Line #29] Define the variable
y_test_predicted
as predicted data forX_test
. - [Line #30] Assign the difference between variables
Y_test
andy_test_predicted
to theresiduals
. - [Line #31] Print the variable
residuals
.
Solución
¡Gracias por tus comentarios!
single
Pregunte a AI
Pregunte a AI
Pregunte lo que quiera o pruebe una de las preguntas sugeridas para comenzar nuestra charla
Resumir este capítulo
Explicar el código en file
Explicar por qué file no resuelve la tarea
Awesome!
Completion rate improved to 4.76
Residuals
Desliza para mostrar el menú
If we look at the plot that shows the dependence of flavanoids on the number of phenols, it will be obvious that the use of linear regression, in this case, was not entirely correct. Moreover, how do we interpret how good our prediction is?
Some points will lie on our constructed line, and some will lie away from it. We can measure the distance between a point and a line along the y-axis. This distance is called the residual or error. The remainder is the difference between the observed value of the target and the predicted value. The closer the residual is to 0, the better our model performs. Let's calculate the residuals and present them as a chart.
12345678residuals = Y_test - y_test_predicted # Visualize the data ax = plt.gca() ax.set_xlabel('total_phenols') ax.set_ylabel('residuals') plt.scatter(X_test, residuals) plt.show()
Output:
Our residuals formed three almost straight lines. This distribution is a sign that the model is not working. Ideally, the remains should be arranged symmetrically and randomly around the horizontal axis. Still, if the residual graph shows some pattern (linear or non-linear), it means that our model is not the best.
Swipe to start coding
Try to find residuals to our previous challenge:
- [Line #29] Define the variable
y_test_predicted
as predicted data forX_test
. - [Line #30] Assign the difference between variables
Y_test
andy_test_predicted
to theresiduals
. - [Line #31] Print the variable
residuals
.
Solución
¡Gracias por tus comentarios!
Awesome!
Completion rate improved to 4.76single