Question

In: Computer Science

What are the advantages of quantile loss over Mean absolute error(MAE) ?

What are the advantages of quantile loss over Mean absolute error(MAE) ?

Solutions

Expert Solution

Quantile loss functions turn out to be useful when we are interested in predicting an interval instead of only point predictions. Prediction interval from least square regression is based on an assumption that residuals (y — y_hat) have constant variance across values of independent variables. We can not trust linear regression models (uses Mean absolute Error) which violate this assumption. We can not also just throw away the idea of fitting linear regression model as baseline by saying that such situations would always be better modeled using non-linear functions or tree based models. This is where quantile loss and quantile regression come to rescue as regression based on quantile loss provides sensible prediction intervals even for residuals with non-constant variance or non-normal distribution.

Understanding the quantile loss function

Quantile-based regression aims to estimate the conditional “quantile” of a response variable given certain values of predictor variables. Quantile loss is actually just an extension of MAE (when quantile is 50th percentile, it’s MAE).

The idea is to choose the quantile value based on whether we want to give more value to positive errors or negative errors. Loss function tries to give different penalties to overestimation and underestimation based on the value of chosen quantile (γ). For example, a quantile loss function of γ = 0.25 gives more penalty to overestimation and tries to keep prediction values a little below median

One advantage of quantile regression, relative to the ordinary least squares regression, is that the quantile regression estimates are more robust against outliers in the response measurements.


Related Solutions

What is the MAPE (mean absolute percentage error) and when would I use it for a...
What is the MAPE (mean absolute percentage error) and when would I use it for a model?
2.   Calculate Mean Absolute Error ( MAD) for the data in question 1 for the three    ...
2.   Calculate Mean Absolute Error ( MAD) for the data in question 1 for the three     methods used. Round MAD to two decimal places. ( 4 marks) Year Revenue 4-Year Moving Average Absolute Error 4 Weighted Moving Average Weights 4,3,2,1 Absolute Error    Exponential    Smoothing          α = 0.6 Absolute Error 2010 75 2011 81 2012 74 2013 79 2014 69 2015 92 2016 73 2017 85 2018 90 2019 73 2020 Forecast What does MAD measures? which of these...
T/f: The mean absolute deviation is more sensitive to large deviations than the mean square error....
T/f: The mean absolute deviation is more sensitive to large deviations than the mean square error. T/f: A smoothing constant of 0.1 will cause an exponential smoothing forecast to react more quickly to a sudden change than a value of 0.3 will. T/f:An advantage of the exponential smoothing forecasting method is that more recent experience is given more weight than less recent experience. T/f: Linear regression can be used to approximate the relationship between independent and dependent variables. T/f:"Forecasting techniques...
what are the Comparative and absolute advantages of brazil from 2000-2019
what are the Comparative and absolute advantages of brazil from 2000-2019
What does absolute advantage mean? How do you calculate absolute advantage? What does comparative advantage mean?...
What does absolute advantage mean? How do you calculate absolute advantage? What does comparative advantage mean? How do you calculate comparative advantage? Does a country with the absolute advantage in the production of a good always have the comparative advantage producing that good? Explain your reasoning.
What does sampling error mean?
What does sampling error mean?
What is the main advantage(s) of quantile regression? How does quantile regression differ from least-squares regression...
What is the main advantage(s) of quantile regression? How does quantile regression differ from least-squares regression (linear regression)? Illustrate your answer with an example.
What is mean, mode, standard deviation, and error?
What is mean, mode, standard deviation, and error?
What are the key advantages, limitations, and sources of error of the titrimetric method of analysis?
What are the key advantages, limitations, and sources of error of the titrimetric method of analysis?
What are the advantages of ICP over flame AA?
What are the advantages of ICP over flame AA?
ADVERTISEMENT
ADVERTISEMENT
ADVERTISEMENT