0:00 examples of supervised machine learning
0:03 supervised learning is one of the most
0:05 common types of machine learning where
0:07 the algorithm is trained on a labeled
0:09 data set in a labeled data set the
0:12 output variable is already known for
0:15 each input variable and the algorithm
0:17 learns to map inputs to outputs based on
0:19 this training data supervised learning
0:22 is used because it relies on labeled
0:25 training data where each image has a
0:27 known category the model learns to
0:30 associate features in the images with
0:32 these predefined
0:34 categories let's consider a supervised
0:36 learning example where we predict
0:38 whether an email is Spam or not spam in
0:42 this case we'll use features associated
0:44 with emails to classify them into these
0:46 two categories features inputs the
0:50 features of the email can include
0:52 various characteristics like the
0:54 presence of certain keywords for example
0:57 when free money the senders email
1:00 address the number of exclamation marks
1:03 in the email the length of the email
1:05 number of words or
1:07 characters whether it contains specific
1:10 patterns like deer name or click here
1:13 label output the label indicates whether
1:17 the email is Spam or not spam or ham
1:20 this is a binary classification problem
1:23 let's illustrate this with a simplified
1:25 diagram on the left side we have two
1:27 examples of email features in in puts
1:30 the first email has certain keywords an
1:33 unknown sender multiple exclamation
1:35 marks a longer length and a specific
1:38 pattern the second email has different
1:41 characteristics in the middle we have
1:43 the model our supervised learning
1:46 algorithm it takes these features as
1:48 input and learns to make predictions on
1:51 the right side we have the output which
1:53 is the label or prediction for the first
1:56 email the model predicts spam and for
1:59 the second email mail it predicts not
2:01 spam or ham supervised learning in this
2:04 case helps us build a model that learns
2:06 the relationship between the email
2:08 features and the labels spam or not spam
2:12 once drained the model can automatically
2:14 classify new unseen emails as spam or
2:17 not spam based on their
2:20 characteristics let's explore another
2:22 supervised learning example where we
2:24 predict whether a student will pass or
2:26 fail an exam based on two features
2:30 features inputs one the number of hours
2:33 a student studied two the number of
2:36 hours a student slept label output
2:40 whether the student passed or failed the
2:42 exam let's create a diagram to
2:44 illustrate this on the left side we have
2:47 two examples of student features inputs
2:51 the first student studied for 4 hours
2:53 and slept for 7 hours the second student
2:57 studied for 2 hours and slept for 5
2:59 hours hours in the middle we have the
3:02 model our supervised learning algorithm
3:05 it takes these features as input and
3:07 learns to predict whether a student will
3:09 pass or fail the exam on the right side
3:12 we have the output which is the label or
3:15 prediction for the first student the
3:18 model predicts parus and for the second
3:20 student it predicts fail supervised
3:23 learning in this example helps us build
3:25 a model that understands the
3:27 relationship between the hours studied
3:29 hour slept and the outcome pars or fail
3:33 once drained the model can predict
3:35 whether a new student will pars or fail
3:37 the exam based on their study and sleep
3:40 patterns another example of supervised
3:43 learning problem predicting whether a
3:46 credit card transaction is fraudulent or
3:48 legitimate based on transaction details
3:51 features inputs one transaction amount
3:55 the amount of money involved in the
3:57 transaction two merchant category
4:00 the type of merchant where the
4:02 transaction occurred for example grocery
4:04 store online retailer free time of day
4:09 the time when the transaction took place
4:11 for example morning afternoon evening
4:14 four location the location where the
4:17 transaction occurred for example City
4:21 Country label output whether the
4:24 transaction is classified as fraudulent
4:26 fraud or legitimate legit here's a
4:30 diagram to illustrate this on the left
4:33 side we have two examples of transaction
4:35 features inputs the first transaction
4:38 has a relatively small amount is from a
4:41 grocery store occurred in the morning
4:43 and took place in New York USA the
4:46 second transaction has a larger amount
4:49 is from an online retailer occurred in
4:51 the evening and the location is unknown
4:54 in the middle we have the model our
4:56 supervised learning algorithm it takes
4:59 these transaction details as input and
5:02 learns to predict whether the
5:03 transaction is classified as fraudulent
5:06 fraud or legitimate legit based on the
5:09 transaction features on the right side
5:12 we have the output which is the label or
5:15 prediction for the first transaction the
5:18 model predicts legit and for the second
5:20 transaction it predicts fraud in this
5:23 example supervised learning helps us
5:25 build a model that understands the
5:27 relationship between transaction details
5:30 and the likelihood of a transaction
5:32 being fraudulent the model can then
5:34 classify in new transactions as
5:36 fraudulent or legitimate based on their
5:39 features thanks for watching if you like
5:42 the tutorial please give it a thumbs up
5:44 and subscribe to our YouTube channel for
5:46 future tutorials
Chrome Extension