Log Loss

Log Loss :


Log loss, also known as cross entropy loss, is a performance metric used in classification tasks. It measures the difference between the predicted probability and the actual outcome.

For example, in a binary classification task where the classes are "positive" and "negative", the predicted probabilities are calculated for each class. If the actual outcome is "positive", the log loss is calculated as:

-log(predicted probability of positive class)

If the predicted probability of the positive class is 0.9, the log loss would be -log(0.9) = 0.10. A lower log loss indicates that the predicted probabilities are closer to the actual outcome.

In another example, if we have a multi-class classification task with 3 classes (A, B, C), the predicted probabilities are calculated for each class. The log loss is calculated as:

-Sum(actual outcome * log(predicted probability of each class))

For example, if the actual outcome is "A" and the predicted probabilities are [0.1, 0.3, 0.6], the log loss would be -(0 * log(0.1) + 1 * log(0.3) + 0 * log(0.6)) = 0.51.

In summary, log loss penalizes heavily for incorrect predictions and rewards for correct predictions. It is commonly used in machine learning competitions as a performance metric, as it is sensitive to probabilities close to 0 and 1, which is important in classification tasks.

Comments

Post a Comment

Popular posts from this blog

Logarithmic transformation

Logarithmic transformation : A logarithmic transformation is a way of changing the scale of an axis on a graph. This can be useful when the data being plotted covers a very wide range of values, making it difficult to see patterns or trends. By using a logarithmic scale, the data can be spread out more evenly and any patterns or trends can be more easily seen. Here are two examples of how a logarithmic transformation can be used: Imagine that you are a scientist studying the growth of a population of bacteria. You take samples of the bacteria at regular intervals and measure the number of bacteria in each sample. Over time, the number of bacteria grows exponentially, so the data you collect looks like a straight line on a graph when plotted on a linear scale. However, it can be difficult to see the details of this growth when the numbers are so large, so you decide to use a logarithmic scale for the y-axis of your graph. This spreads out the data and makes it easier to see the growth o...
Read more

Mean squares

Mean squares : The term "mean squares" refers to a statistical calculation that is used to determine the difference between two sets of data. This calculation is typically used in hypothesis testing, where it helps researchers to determine whether there is a significant difference between the two sets of data. To understand how mean squares works, let's consider the following example: Suppose that a researcher is interested in studying the relationship between the amount of time that students spend studying and their grades on a particular test. The researcher collects data from a sample of students and finds that the average grade for students who studied for more than 4 hours is 87, while the average grade for students who studied for less than 4 hours is 73. The researcher can use mean squares to determine whether there is a significant difference between the two groups of students. To do this, the researcher would first calculate the mean square for each group of stud...
Read more

Markov chain

Markov chain : A Markov chain is a mathematical system that undergoes transitions from one state to another according to certain probabilistic rules. The defining characteristic of a Markov chain is that no matter how the system arrived at its current state, the possible future states are fixed. This is known as the "memoryless" property. One example of a Markov chain is a model for the weather. Let's say we have three states: sunny, cloudy, and rainy. The probability of transitioning from one state to another is fixed and known. For instance, the probability of transitioning from sunny to cloudy may be 0.3, while the probability of transitioning from cloudy to rainy may be 0.7. If we start in the sunny state, we can use the probabilities to calculate the likelihood of transitioning to the other states. For example, the probability of transitioning to the cloudy state is 0.3, and the probability of transitioning to the rainy state is 0. If we stay in the sunny state for s...
Read more