Classification metrics optimization II

Loading...

Из курса от партнера National Research University Higher School of Economics

How to Win a Data Science Competition: Learn from Top Kagglers

462 оценки

Explore Related Videos

At Coursera, you will find the best lectures in the world. Here are some of our personalized recommendations for you

National Research University Higher School of Economics

How to Win a Data Science Competition: Learn from Top Kagglers

462 оценки

Курс 2 из 7 — Specialization Advanced Machine Learning

If you want to break into competitive data science, then this course is for you! Participating in predictive modelling competitions can help you gain practical experience, improve and harness your data modelling skills in various domains such as credit, insurance, marketing, natural language processing, sales’ forecasting and computer vision to name a few. At the same time you get to do it in a competitive context against thousands of participants where each one tries to build the most predictive algorithm. Pushing each other to the limit can result in better performance and smaller prediction errors. Being able to achieve high ranks consistently can help you accelerate your career in data science. In this course, you will learn to analyse and solve competitively such predictive modelling tasks. When you finish this class, you will: - Understand how to solve predictive modelling competitions efficiently and learn which of the skills obtained can be applicable to real-world tasks. - Learn how to preprocess the data and generate new features from various sources such as text and images. - Be taught advanced feature engineering techniques like generating mean-encodings, using aggregated statistical measures or finding nearest neighbors as a means to improve your predictions. - Be able to form reliable cross validation methodologies that help you benchmark your solutions and avoid overfitting or underfitting when tested with unobserved (test) data. - Gain experience of analysing and interpreting the data. You will become aware of inconsistencies, high noise levels, errors and other data-related issues such as leakages and you will learn how to overcome them. - Acquire knowledge of different algorithms and learn how to efficiently tune their hyperparameters and achieve top performance. - Master the art of combining different machine learning models and learn how to ensemble. - Get exposed to past (winning) solutions and codes and learn how to read them. Disclaimer : This is not a machine learning course in the general sense. This course will teach you how to get high-rank solutions against thousands of competitors with focus on practical usage of machine learning methods rather than the theoretical underpinnings behind them. Prerequisites: - Python: work with DataFrames in pandas, plot figures in matplotlib, import and train models from scikit-learn, XGBoost, LightGBM. - Machine Learning: basic understanding of linear models, K-NN, random forest, gradient boosting and neural networks.

Из урока

Metrics Optimization

This week we will first study another component of the competitions: the evaluation metrics. We will recap the most prominent ones and then see, how we can efficiently optimize a metric given in a competition.

Regression metrics review I14:21

Regression metrics review II8:29

Classification metrics review20:43

General approaches for metrics optimization6:29

Regression metrics optimization10:05

Classification metrics optimization I7:33

Classification metrics optimization II6:58

Познакомьтесь с преподавателями

Dmitry Ulyanov
Visiting lecturer
HSE Faculty of Computer Science
Alexander Guschin
Visiting lecturer at HSE, Lecturer at MIPT
HSE Faculty of Computer Science
Mikhail Trofimov
Visiting lecturer
HSE Faculty of Computer Science
Dmitry Altukhov
Visiting lecturer
HSE Faculty of Computer Science
Marios Michailidis
Research Data Scientist
H2O.ai

So in the previous video,

we've discussed Logloss and Accuracy.

In this video we'll discuss Area Under Curve,

AUC, and (Quadratic weighted) Kappa.

Let's start with AUC.

Although the loss function of AUC has zero gradients almost everywhere,

exactly as accuracy loss,

there exists an algorithm to optimize AUC with gradient-based methods,

and some models implement this algorithm.

So we can use it by setting the right parameters.

I will give you an idea about this method without

much details as there is more than one way to implement it.

Recall that originally, classification task is usually solved at the level of objects.

We want to assign 0 to red objects,

and 1 to the green ones.

But we do it independently for each object,

and so our loss is pointwise.

We compute it for each object individually,

and sum or average the losses for all the objects to get a total loss.

Now, recall that AUC is the probability

of a pair of the objects to be ordered in the right way.

So ideally, we want

predictions Y^ for the green objects to be larger than for the red ones.

So, instead of working with single objects,

we should work with pairs of objects.

And instead of using pointwise loss,

we should use pairwise loss.

A pairwise loss takes predictions and labels

for a pair of objects and computes their loss.

Ideally, the loss would be zero when the ordering is correct,

and greater than zero when the ordering is not correct, incorrect.

But in practice, different loss functions can be used.

For example, we can use logloss.

We may think that the target for this pairwise loss is always one,

red minus green should be one.

That is why there is only one term in logloss objective instead of two.

The prob function in the formula is needed to

make sure that the difference between the predictions is still in the 0,1 range,

and I use it here just for the sake of simplicity.

Well, basically, XGBoost, LightGBM have pairwise loss we've discussed implemented.

It is straightforward to implement in any neural net library,

and for sure, you can find implementations on GitHub.

I should say that in practice,

most people still use logloss as an optimization loss without any more post processing.

I personally observed XGBoost learned with loglosst to give

comparable AUC score to the one learned with pairwise loss. All right.

Now, let's move to the last topic to discuss.

It is Quadratic weighted Kappa metric. There are two methods.

One is very common and very easy,

the second is not that common and will require you to implement

a custom loss function for either XGBoost or neural net.

But we've already implemented it for XGBoost,

so you will be able to find the implementation in the reading materials.

But let's start with the simple one.

Recall that we're solving

an ordered classification problem and our labels can be found of us integer ratings,

say from one to five.

The task is classification as we cannot output,

for example, 4.5 as an answer.

But anyway, we can treat it as a regression problem, and then somehow,

post-process the predictions and convert them to integer ratings.

And actually quadratic weights make Kappa as somehow similar to regression with MSE loss.

If we allow our predictions to take values between the labels,

that is relax the predictions.

But in fact, it is different to MSE.

So if relaxed, Kappa would be

one minus MSE divided by something that really depends on the predictions.

And it looks like everyone's logic is, well,

there is MSE in the denominator,

we can optimize it,

and let's don't care about denominator.

Well, of course it's not correct way to do it,

but it turns out to be useful in practice.

But anyway, MSE gives us flat values instead of integers.

So now, we need somehow to convert them into integers.

And the straightforward way would be to do rounding all the predictions.

But we can think about rounding as of applying a threshold.

Like if the value is greater than 3.5 and less than 4.5, then output 3.

But then we can ask ourselves a question,

why do we use exactly those thresholds? Let's tune them.

And again, it's just straightforward,

it can be easily done with grid search.

So to summarize, we need to fit MSE loss to

our data and then find appropriate thresholds.

Finally, there is a paper which

suggests a way to relax classification problem to regression,

but it deals with this- hard to deal with part in denominator that we had.

I will not get into the details here,

but it's clearly written and easy to understand paper,

so I really encourage you to read it.

And more, you can find loss implementation in the reading materials,

and just use it if you don't want to read the paper.

Finally, we finished this lesson.

We've discussed that evaluation or target metric is how all submissions are scored.

We've discussed the difference between target metric and optimization loss.

Optimization loss is what our model optimizes,

and it is not always the same as target metric that we want to optimize.

Sometimes, we only can set our model to optimize completely different to target metric.

But later, we usually try to post-process the predictions

to make them better fit target metric.

We've discussed intuition behind

different metrics for regression and classification tasks,

and saw how to efficiently optimize different metrics.

I hope you've enjoyed this lesson,

and see you later.

Ознакомьтесь с нашим каталогом

Присоединяйтесь бесплатно и получайте персонализированные рекомендации, обновления и предложения.

Coursera делает лучшее в мире образование доступным каждому, предлагая онлайн-курсы от ведущих университетов и организаций.

© Coursera Inc., 2019 Все права защищены.

Загрузить из App Store

Загрузить в Google Play