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
Frisian
Ga
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 (Soranรฎ)
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
[Autogenerated] in this clip, Let's first
more to discussing feature combination.
Now it's quite possible that the raw data
that you're working with contains very
granular information, which doesn't have
much predictive power. Feature combination
may involve aggregating and bringing
features together to get a future with
more political power. Now you might find
it in the real bold. Some features
naturally work better when they're
considered together. Ah, feature by itself
may not contain much information, but then
considered in conjunction with another
feature. The teachers and combination
might contain information that is relevant
to your model. It's quite possible that
the original feature might be to row or
two granular. Bringing features together
can help improve the predictive power of
features. Let's say you're building a
machine learning model tow. Predict
traffic patterns in a city. Let's say the
city is a bang lord. Now you might get
information from the day off the beat that
it is. You might also get information from
the time off the But when taken in
conjunction, when you use a future cross
day off the week plus time off the day,
you might get a resulting feature that has
more predictive power If you're looking at
traffic at Friday evening at 60 m, you
know it's going to be terrible. But if
you're looking at the same time, 60 M on a
Sunday traffic is quite likely not as bad.
Let's say you want to combine features
together to predict temperature. You can
take into account the current season
whether it's spring, summer, fall or
winter. You can also take into account the
time off the but when taken together, you
might find that the feature combination is
more than the sum of the parts. And
finally, let's move on to the last
component that we'll discuss with in
future engineering. At this dimensionality
reduction. When you're working with data
in the real world, you will find that a
common problem toe have is that you have
too much data. This is a curse and not a
blessing, and it's often referred to as
the cost of dimensionality. This is where
he would apply pre processing algorithms
to reduce the complexity off raw features,
and the specific aim off these algorithms
is to reduce the number of input features
so you have fewer features to work with
having too many features to work within
your data is referred to us the cost of
dimensionality, and it leads to several
problems. You have problems visualizing
your data. You encounter problems during
training as Celeste hearing prediction.
When you work with higher dimensionality
data machine learning models find it hard
to find patterns within your data leading
toe poor quality models over fitted
models. Forfeited models are those that
perform well in training but poorly in the
real world. In production, dimensionality
reduction explicitly aims to solve the
coast of dimensionality while preserving
as much information as possible from the
underlying features. You don't want to
lose too much information. Dimensionality
reduction is a form of unsupervised
learning. You're working with an unlabeled
corpus of data based on the kind of data
that you're working with. There are many
different techniques that you can use for
dimensionality reduction. When you're
working with linear data, you can choose
principle components analysis, which
involves Lee orienting your original data
so that their projected along newer,
better axes. If you're working with non
linear data, you can apply manifold
learning techniques. This involved
unrolling complex forms off Dayton and
higher access toe express data in a
simpler form with lower dimensionality.
Manafort Island techniques are similar toe
unrolling a carpet so that it's
represented in two dimensions. Leading
semantic analysis is a topic modeling and
dimensionality reduction technique that
you can use to work with text data if
you're working with images auto, including
confined, efficient, lower dimensionality representation for your images.
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.