方差偏差权衡_偏差偏差权衡：快速介绍
方差偏差權(quán)衡
The bias-variance tradeoff is one of the most important but overlooked and misunderstood topics in ML. So, here we want to cover the topic in a simple and short way as possible.
偏差-方差折衷是機(jī)器學(xué)習(xí)中最重要但被忽視和誤解的主題之一。 因此，在這里我們想以一種簡(jiǎn)單而簡(jiǎn)短的方式涵蓋這個(gè)主題。
Let’s start with basics and see why it is important and how this concept is to be used. We want to keep this crisp so we’ll talk in pointers at times. By the end of this, you would know:
讓我們從基礎(chǔ)開始，看看為什么它很重要以及如何使用此概念。 我們希望保持清晰，因此我們有時(shí)會(huì)談指針。 到此為止，您將知道：
- Bias 偏壓
- Variance 方差
- Their relationship 他們的關(guān)系
- importance of their tradeoff 權(quán)衡的重要性
- how to analyze the model condition and take necessary steps 如何分析模型條件并采取必要步驟
So what this Bias- Variance tradeoff exactly has to do with performance?
那么，這種偏差與偏差的權(quán)衡與性能究竟有什么關(guān)系呢？
You build a model. The model doesn’t perform well. You want to improve the performance but don’t know where to start.
您建立模型。 該模型的效果不佳。 您想提高性能，但不知道從哪里開始。
A diagnosis is important as it pin-points the areas of improvement. You need to clearly identify the components which are leading to a poor model.
診斷很重要，因?yàn)樗梢灾赋鲂枰倪M(jìn)的地方。 您需要清楚地識(shí)別導(dǎo)致模型不良的組件。
Issue: Bad model performance
問題 ：模型性能不佳
Focus area for the fix: Prediction error
修復(fù)的重點(diǎn)領(lǐng)域：預(yù)測(cè)錯(cuò)誤
Before jumping to the topic, just know this.
在跳到主題之前，請(qǐng)先知道這一點(diǎn)。
Total Error = Bias^2 + Variance + Irreducible error
Total Error = Bias^2 + Variance + Irreducible error
a. Total error = Prediction error that we are trying to minimize
一個(gè)。 總誤差 =我們?cè)噲D最小化的預(yù)測(cè)誤差
b. Bias error = Difference between the average prediction of model and the correct prediction
b。 偏差誤差 =模型的平均預(yù)測(cè)與正確預(yù)測(cè)之間的差異
c. Variance = Variability of a model prediction for a given data point (difference in results for the same data point if training data is changed).
C。 方差 =給定數(shù)據(jù)點(diǎn)的模型預(yù)測(cè)的方差 (如果更改訓(xùn)練數(shù)據(jù)，則同一數(shù)據(jù)點(diǎn)的結(jié)果差異)。
d. Irreducible error = It is the inherent error of data which is caused by the distribution of data and other specification. It is just the way the data is and basically nothing can be done about it.
d。 不可減少的錯(cuò)誤 =這是由數(shù)據(jù)的分布和其他規(guī)范引起的數(shù)據(jù)固有的錯(cuò)誤。 這只是數(shù)據(jù)的方式，基本上它無法做任何事情。
Okay, these are formal definitions. How to visualize them and understand them in normal terms.
好的，這些是正式的定義。 如何可視化它們并以常規(guī)術(shù)語理解它們。
Goal — Low Bias, Low Variance
目標(biāo)—低偏見，低方差
Fig 1: Bias-Variance representation圖1：偏差-方差表示Let’s see each of the possible combinations and understand each of them practically with the above representation.
讓我們看一下每種可能的組合，并通過上述表示實(shí)際理解它們。
a. High Bias, high Variance: Worst Case- Results not close to the target(High Bias) and not even consistent in any direction(High variance).
一個(gè)。 高偏差，高方差： 最壞的情況-結(jié)果不接近目標(biāo)(高偏差)，甚至在任何方向上都不一致(高偏差)。
b. High Bias, low variance: Results not close to the target (High Bias) but consistent in one direction(Low Variance).
b。 高偏差，低方差：結(jié)果不接近目標(biāo)(高偏差)，但在一個(gè)方向上一致(低方差)。
c. Low Bias, high Variance: Results close to the target (Low Bias) but not consistent around the target(High Variance).
C。 低偏差，高方差：結(jié)果接近目標(biāo)(低偏差)，但在目標(biāo)周圍(高方差)不一致。
d.Low Bias, low variance: Best Case- Results close to the target (Low Bias) and consistent around the target(Low Variance).
d。 低偏差，低方差： 最佳情況-結(jié)果接近目標(biāo)(低偏差)，并且在目標(biāo)附近(低方差)保持一致。
Now the question is why it is a tradeoff. Why not simply go and get low bias low variance. This is because of the way bias and variance are related, each comes at the cost of other. When you try to improve one, the other gets worse. Like if you cook on low flame, it takes forever. You increase the flame, food starts burning. You have to find a point where both are balanced.
現(xiàn)在的問題是，為什么要進(jìn)行權(quán)衡。 為什么不簡(jiǎn)單地去獲得低偏差低方差。 這是因?yàn)槠詈头讲钪g是相關(guān)的，每個(gè)都以其他為代價(jià)。 當(dāng)您嘗試改善一個(gè)時(shí)，另一個(gè)會(huì)變得更糟。 就像您在低火上烹飪一樣，它需要永遠(yuǎn)。 您增加火焰，食物開始燃燒。 您必須找到一個(gè)平衡點(diǎn)。
Fig. 2圖2Ideal model: Learns the underlying patterns in training data just optimally and creates a generalized algorithm that can work with similar unseen data as well.
理想模型：以最佳方式學(xué)習(xí)訓(xùn)練數(shù)據(jù)中的基礎(chǔ)模式，并創(chuàng)建一種通用算法，該算法也可以處理相似的看不見的數(shù)據(jù)。
Overfitting: The model makes a very highly fitting algorithm tailored for the training data specifically. Thus, it cannot stand variations that come with unseen data.
過度擬合：模型針對(duì)訓(xùn)練數(shù)據(jù)制定了非常適合的算法。 因此，它無法忍受看不見的數(shù)據(jù)帶來的變化。
An overfitting model can be understood as a “Frog in the well” who became too comfortable in the present scenario(training data) but its present understanding won’t help to survive a different surrounding(test data).
過度擬合模型可以理解為“井中的青蛙”，他在當(dāng)前場(chǎng)景(訓(xùn)練數(shù)據(jù))中變得太自在了，但其目前的理解無助于在不同的環(huán)境中生存(測(cè)試數(shù)據(jù))。
Underfitting: The model makes a very loose-fitting algorithm that can’t even work for the training data as it couldn’t learn the patterns as it oversimplified everything. Thus it cannot give correct answers.
欠擬合：模型提出了一種非常寬松的算法，該算法甚至不能用于訓(xùn)練數(shù)據(jù)，因?yàn)樗^于簡(jiǎn)化了所有操作，因此無法學(xué)習(xí)模式。 因此，它不能給出正確的答案。
An underfitting model is a person who thinks he learned a skill by just taking the intro session and learning buzz words or he became a cricket player just because he knows how to hit a ball.
不稱職的模特是一個(gè)人，他認(rèn)為自己只是通過參加入門課程并學(xué)習(xí)流行語來學(xué)習(xí)技能，或者僅僅因?yàn)樗廊绾螕羟蚨蔀榘迩蜻\(yùn)動(dòng)員。
You can read the detailed explanation below:
您可以閱讀以下詳細(xì)說明：
https://medium.com/analytics-vidhya/understanding-how-machine-learning-is-just-like-the-human-learning-process-801a0bca3e56
https://medium.com/analytics-vidhya/understanding-how-machine-learning-is-just-like-the-human-learning-process-801a0bca3e56
The goal was to build a model that gives-
目的是建立一個(gè)模型，使-
Right results most of the times.
在大多數(shù)情況下，正確的結(jié)果。
Models with Overfitting have high variance and ones with Underfitting have a high bias.
具有過度擬合的模型具有較高的方差，而具有欠擬合的模型具有較高的偏差。
What do I keep in mind regarding these to solve them in real-time?
我要牢記這些以實(shí)時(shí)解決這些問題？
- Identify whether your model suffers from overfitting or underfitting. Use the train-test accuracy of the model for this. 確定您的模型是過度擬合還是擬合不足。 為此，請(qǐng)使用模型的訓(xùn)練測(cè)試精度。
- Take measures as follows once the issue is identified. 一旦發(fā)現(xiàn)問題，請(qǐng)采取以下措施。
a. Problem: High Variance(This will be solved the way overfitting is solved)
一個(gè)。 問題：高方差(將通過解決過度擬合的方式解決)
Let’s see each solution and how exactly it is solving the issue.
讓我們看看每種解決方案以及它如何解決問題。
Add more training data: You have learned very data specific. Here’s more data for increasing your general understanding so that it is no longer data specific.
添加更多培訓(xùn)數(shù)據(jù) ：您已經(jīng)學(xué)到了非常具體的數(shù)據(jù)。 這里有更多數(shù)據(jù)可用于增強(qiáng)您的一般理解，從而不再是特定于數(shù)據(jù)的數(shù)據(jù)。
Data augmentation: I don’t have much data. Let me modify current data to create more variations and present them to you for your better understanding.
數(shù)據(jù)擴(kuò)充 ：我沒有太多數(shù)據(jù)。 讓我修改當(dāng)前數(shù)據(jù)以創(chuàng)建更多變體，然后將其呈現(xiàn)給您，以使您更好地理解。
Reduce the complexity of the model: You have learned unnecessary stuff. These specific details are not required. Retain only what can be applied everywhere and let go of rest to simplify.
降低模型的復(fù)雜性 ：您已經(jīng)學(xué)到了不必要的東西。 這些特定的細(xì)節(jié)不是必需的。 僅保留可以在任何地方應(yīng)用的內(nèi)容，并放手休息以簡(jiǎn)化操作。
Bagging(stands for Bootstrap Aggregating): You are giving different answers every time I change the training data a little. Let me randomly sample the data and give to you all the samples. You create predictors and train on each sample and get all the different results you can. Put together all learning by aggregating all the results and give me one final answer which will remain consistent.
Bagging(代表B ootstrap Agg的注冊(cè)) ：每次我稍微改變訓(xùn)練數(shù)據(jù)，您都會(huì)給出不同的答案。 讓我隨機(jī)抽樣數(shù)據(jù)，然后給您所有樣本。 您可以創(chuàng)建預(yù)測(cè)變量并對(duì)每個(gè)樣本進(jìn)行訓(xùn)練，并獲得所有不同的結(jié)果。 通過匯總所有結(jié)果來匯總所有學(xué)習(xí)內(nèi)容，并給我一個(gè)最終答案，它將保持一致。
Note: The different predictors need to have minimum correlation so that they make “different errors”(not to be confused with the model, we have 1 model having different predictors that gives results of different samples).
注意：不同的預(yù)測(cè)變量需要具有最小的相關(guān)性，以使它們產(chǎn)生“不同的誤差”(不要與模型混淆，我們有1個(gè)具有不同預(yù)測(cè)變量的模型可以得出不同樣本的結(jié)果)。
b. Problem: High Bias(This will be solved the way underfitting is solved).
b。 問題 ：高偏差(這將通過解決欠擬合的方式解決)。
Add features: You gave a result that Person A won’t be able to repay the loan because he is old(feature). You are saying this because an old Person B couldn’t repay it. But you also need to see their annual income, past history, etc(other features) and then decide.
添加功能：您得出的結(jié)果是，人A由于年齡大(功能)而無法償還貸款。 您之所以這樣說，是因?yàn)槔先薆無法償還。 但是，您還需要查看他們的年收入，過去的歷史記錄等(其他功能)，然后做出決定。
Use a model with higher complexity: We need to replace you with someone who can understand the relation between different parts of data and how they work together better than you.
使用具有更高復(fù)雜性的模型：我們需要以能夠理解數(shù)據(jù)不同部分之間的關(guān)??系以及它們?nèi)绾胃玫貐f(xié)同工作的人來代替您。
Boosting: I don’t trust you. You create predictors and ask them each to answer. We’ll ask each predictor about the logic they used to get their partially right answers. Whenever we get some part right, we’ll add that logic in the rule. Each one will have their shortcoming but together, they will cover up for each other. They’ll work as a team to finally create a well-fitting complex rule.
提振：我不相信你。 您創(chuàng)建預(yù)測(cè)變量，并要求它們各自回答。 我們將詢問每個(gè)預(yù)測(cè)變量有關(guān)他們用來獲得部分正確答案的邏輯。 每當(dāng)我們得到正確的部分時(shí)，我們都會(huì)在規(guī)則中添加該邏輯。 每個(gè)人都會(huì)有自己的缺點(diǎn)，但是他們會(huì)互相掩蓋。 他們將作為一個(gè)團(tuán)隊(duì)工作，以最終創(chuàng)建一個(gè)合適的復(fù)雜規(guī)則。
Note: The team of weak learners should have a minimum correlation between them, otherwise everyone would have the right answers for the same sections and some sections will be left answered incorrectly.
注意：弱學(xué)習(xí)者團(tuán)隊(duì)之間的相關(guān)性應(yīng)最低，否則每個(gè)人對(duì)相同部分的回答都是正確的，而某些部分的回答將不正確。
Hope this helped to understand the topic and gave the understanding to leverage the concept as well.
希望這有助于理解該主題，并給予理解以利用該概念。
Let us know your feedback. Thanks for reading!
讓我們知道您的反饋。 謝謝閱讀！
Sources:
資料來源：
Fig 1, Fig 2: http://scott.fortmann-roe.com/docs/BiasVariance.html
圖1，圖2： http ： //scott.fortmann-roe.com/docs/BiasVariance.htm
翻譯自: https://medium.com/analytics-vidhya/bias-variance-tradeoff-a-quick-introduction-a4b55e56fa24
方差偏差權(quán)衡
總結(jié)
以上是生活随笔為你收集整理的方差偏差权衡_偏差偏差权衡：快速介绍的全部?jī)?nèi)容，希望文章能夠幫你解決所遇到的問題。
- 上一篇： 梦到被绿蛇追着是什么意思
- 下一篇： 分节符缩写p_p值的缩写是什么？