subtitlecat.com

All language subtitles for 06 - Components of Feature Engineering.en

Afrikaans

Akan

Albanian

Amharic

Arabic

Armenian

Azerbaijani

Basque

Belarusian

Bemba

Bengali

Bihari

Bosnian

Breton

Bulgarian

Cambodian

Catalan

Cebuano

Cherokee

Chichewa

Chinese (Simplified)

Chinese (Traditional)

Corsican

Croatian

Czech

Danish

Dutch

English

Esperanto

Estonian

Ewe

Faroese

Filipino

Finnish

French Download

Frisian

Galician

Georgian

German

Greek

Guarani

Gujarati

Haitian Creole

Hausa

Hawaiian

Hebrew

Hindi

Hmong

Hungarian

Icelandic

Igbo

Indonesian

Interlingua

Irish

Italian

Japanese

Javanese

Kannada

Kazakh

Kinyarwanda

Kirundi

Kongo

Korean

Krio (Sierra Leone)

Kurdish

Kurdish (Soranî)

Kyrgyz

Laothian

Latin

Latvian

Lingala

Lithuanian

Lozi

Luganda

Luo

Luxembourgish

Macedonian

Malagasy

Malay

Malayalam

Maltese

Maori

Marathi

Mauritian Creole

Moldavian

Mongolian

Myanmar (Burmese)

Montenegrin

Nepali

Nigerian Pidgin

Northern Sotho

Norwegian

Norwegian (Nynorsk)

Occitan

Oriya

Oromo

Pashto

Persian

Polish

Portuguese (Brazil)

Portuguese (Portugal)

Punjabi

Quechua

Romanian

Romansh

Runyakitara

Russian

Samoan

Scots Gaelic

Serbian

Serbo-Croatian

Sesotho

Setswana

Seychellois Creole

Shona

Sindhi

Sinhalese

Slovak

Slovenian

Somali

Spanish

Spanish (Latin American)

Sundanese

Swahili

Swedish

Tajik

Tamil

Tatar

Telugu

Thai

Tigrinya

Tonga

Tshiluba

Tumbuka

Turkish

Turkmen

Twi

Uighur

Ukrainian

Urdu

Uzbek

Vietnamese

Welsh

Wolof

Xhosa

Yiddish

Yoruba

Zulu

Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated: 0 00:00:01,040 --> 00:00:02,200 [Autogenerated] all of our discussion so 1 00:00:02,200 --> 00:00:03,940 far should have emphasized to you the 2 00:00:03,940 --> 00:00:06,000 importance off the features that you used 3 00:00:06,000 --> 00:00:07,839 to train your machine learning models. 4 00:00:07,839 --> 00:00:09,560 When we spoke off the basic machine 5 00:00:09,560 --> 00:00:11,519 learning workflow in the previous clip, we 6 00:00:11,519 --> 00:00:14,000 discussed the importance off the 1st 3 7 00:00:14,000 --> 00:00:16,829 steps. Selecting and extracting the right 8 00:00:16,829 --> 00:00:19,269 features to train your model. These are 9 00:00:19,269 --> 00:00:22,219 onerous, and I am consuming on. All of 10 00:00:22,219 --> 00:00:25,219 these encompass feature engineering. What 11 00:00:25,219 --> 00:00:27,839 exactly is featured engineering? Well, 12 00:00:27,839 --> 00:00:29,789 feature engineering basically involves 13 00:00:29,789 --> 00:00:31,800 working with your features. Engineering 14 00:00:31,800 --> 00:00:34,070 your features so that you get the best out 15 00:00:34,070 --> 00:00:36,060 off your machine learning Marty. And this 16 00:00:36,060 --> 00:00:38,390 can be anything. Feature. Engineering 17 00:00:38,390 --> 00:00:40,649 isn't just one technique or a set off 18 00:00:40,649 --> 00:00:42,719 techniques or a class of techniques. It's 19 00:00:42,719 --> 00:00:45,899 block and tackle work. You're inevitably 20 00:00:45,899 --> 00:00:47,780 barking and improving the features off 21 00:00:47,780 --> 00:00:50,600 your model. Also featured engineering 22 00:00:50,600 --> 00:00:53,200 tends to be bespoke. It's specific to the 23 00:00:53,200 --> 00:00:55,109 problem that you're working on and the 24 00:00:55,109 --> 00:00:57,259 data that you have to work with. There are 25 00:00:57,259 --> 00:01:00,000 no set of techniques that apply to all 26 00:01:00,000 --> 00:01:02,909 classified files or regresses. You can't 27 00:01:02,909 --> 00:01:04,799 say that this kind of future engineering 28 00:01:04,799 --> 00:01:07,079 works best for noodle networks. Where is 29 00:01:07,079 --> 00:01:09,000 this other kind of future? Engineering 30 00:01:09,000 --> 00:01:11,569 works were traditionally male models. The 31 00:01:11,569 --> 00:01:13,269 ideal and principles off feature 32 00:01:13,269 --> 00:01:16,030 engineering lie along a continuum between 33 00:01:16,030 --> 00:01:18,519 art and science. It's not quite art. It's 34 00:01:18,519 --> 00:01:20,459 not quite signs. It's more just 35 00:01:20,459 --> 00:01:22,950 engineering, where you basically bring 36 00:01:22,950 --> 00:01:25,349 your features together in a form that 37 00:01:25,349 --> 00:01:28,200 build robust models. There is no one size 38 00:01:28,200 --> 00:01:31,569 fits all imagined. Future engineering as a 39 00:01:31,569 --> 00:01:34,219 very broad umbrella with encompasses a 40 00:01:34,219 --> 00:01:35,709 number of different techniques. This 41 00:01:35,709 --> 00:01:37,879 involves feature selection using 42 00:01:37,879 --> 00:01:40,060 techniques to select the most relevant 43 00:01:40,060 --> 00:01:42,500 features for your model. Feature. 44 00:01:42,500 --> 00:01:45,069 Engineering also encompasses feature 45 00:01:45,069 --> 00:01:47,370 learning where you can use supervised on 46 00:01:47,370 --> 00:01:49,840 unsupervised techniques to learn late and 47 00:01:49,840 --> 00:01:52,390 features that exists in your data. Feature 48 00:01:52,390 --> 00:01:54,760 engineering includes feature extraction as 49 00:01:54,760 --> 00:01:57,420 well. Feature extraction involves the 50 00:01:57,420 --> 00:01:59,909 transformation or reorientation off your 51 00:01:59,909 --> 00:02:02,390 input features into fundamentally 52 00:02:02,390 --> 00:02:04,629 transformed derived features, which are 53 00:02:04,629 --> 00:02:06,879 often unrecognizable and cannot be 54 00:02:06,879 --> 00:02:09,610 interpreted. Feature engineering also 55 00:02:09,610 --> 00:02:11,889 include FEATURE combination. You might 56 00:02:11,889 --> 00:02:14,669 have raw granular features in your data. 57 00:02:14,669 --> 00:02:17,139 You might combine these features together 58 00:02:17,139 --> 00:02:19,150 to get a feature that is more meaningful 59 00:02:19,150 --> 00:02:21,669 and has more predictive power. And 60 00:02:21,669 --> 00:02:23,710 finally, feature engineering includes 61 00:02:23,710 --> 00:02:25,840 dimensionality reduction as well. 62 00:02:25,840 --> 00:02:28,150 Dimensionality reduction involves reducing 63 00:02:28,150 --> 00:02:30,360 the complexity off your input data. It 64 00:02:30,360 --> 00:02:32,770 also involves re orienting your features 65 00:02:32,770 --> 00:02:36,000 along new axes, which better represent your data 5112