Spam Detection:

Spam via emails, SMS or
any other medium is very dangerous as it may cause loss of money or loss of
other important belongings or leaking of important data. Spam is very dangerous
to any individual or organization due to its nature of loss of capita and
information leaking. Some spams may even inject malware into the system only by
just opening them. Spam detection works by previously reported or flag(ed) or
by the system itself by analysing the contents.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

 

How will you choose the
right algorithm?

Machine learning
algorithm is categorized in three types they are Classification, Regression and
Clustering. An email may be either spam or will be not, so classification is
the best choice for this as, it should fit in either class -spam or not spam.

 

Write in detail about the different steps that you
will perform to develop the machine learning algorithm.

Collect Data: Data is collected from previous mails from email
clients that are already sorted.

Analyze Data: This data is
analyzed to understand the information available like what makes an email to be
spam, etc.

Prepare Data: Having
a proper dataset with proper emails as some incomplete emails without (any
meaning) must be removed or information must be added so as to improve the
dataset for the purpose it is to be used for. Only data which will help to
detect whether it is spam or normal email must be used i.e. useless data must
be removed and take data which would improve results.

Train and Test: This
data will be trained using 80% of training set and tested using 20% of the
training set.

 

 

Automated Braking:

Automated Braking may be
part of a whole system (Automated driving) or an independent part where only
braking is automated (Brake Assist).

 

How will you choose the
right algorithm?

Braking is not like
either full braking (100) and coming to stop or no braking (0) at all, Braking
have multiple values which may be applied according to the present scenario.
Hence regression is the best machine learning algorithm for automated braking.

Here the braking system
will use input sensors for detecting signals where full brake will be applied
to bring vehicle to standstill, while for other cases where an obstacle comes,
say like speed braker (road bumps) where the vehicle should be brought under a
certain speed only, etc.

 

Write in detail about the different steps that you
will perform to develop the machine learning algorithm.

Collect Data: Data is collected from manufacturer about braking,
etc. for accuracy.

Analyze Data: Study parameters and their relations that affect
braking of a vehicle like speed, type of brake, braking power, type of surface(road),
type of tyre, friction value, etc.

Prepare Data: Dataset in the form of regression tree (or CART) where
various factor mentioned above with value ranges and the target(output) as how
much braking would be needed. As braking power is not calculated in time since
braking will eventually led to stopping of the vehicle which would be not
useful on road bumps, so instead output would be required in terms of speed
like for speed bumps 25kmph, for red signals and other emergency braking etc. 0
kmph, etc.

Train and Test: This data will be trained using 80% of training set
and tested using 20% of the training set.

 

 

Stock Market Prediction:

Stock market prediction is very difficult job as wrong
prediction may lead to loss of capital. As stock market depends (say of a
particular company/organization) on its performance- current and previous,
decisions taken and their effects (like cancelling a product can also lead to
low stock price), etc. as this factor influence investors to either invest or
withdraw shares.

 

How will you choose the
right algorithm?

As stock market is dependent on various factors and
predicting if price per share would rise or fall and by how much would require
regression due to numeric prediction.

 

Write in detail about the different steps that you
will perform to develop the machine learning algorithm.

Collect Data: By mapping the stock price by the date and time.

Analyze Data: Analyzing the factors that cause people to invest or
withdraw that results in shares to rise or fall.

Prepare Data: Dataset will be then cleaned by removing incomplete data
or adding missing information where there is missing price on a particular date
and time, etc.

Train and Test: This data will be trained using 80% of training set
and tested using 20% of the training set.

 

 

Recommendations:

Recommendations anywhere online depends on previously
searched or visited content. They show same or similar content with prices and
discounts or other benefits.

 

How will you choose the
right algorithm?

Recommendations track our previous sessions and
suggest accordingly, it may recommend similar product or product which is
likely to be purchased. Recommendation may be from same category, say book from
sci-fi genre so recommendation may give other sci-fi books as well as books
from another genre too. So, suggesting from other genre would mean clustering
as it tries to match (checks for similarity) for fitting data.

 

Write in detail about the different steps that you
will perform to develop the machine learning algorithm.

Collect Data: By similar products together, history i.e. what user
check after checking a product.

Analyze Data: Learning how people would most likely check the
recommendation from the data provided.

Prepare Data: Cleaning the data by selecting data which is more
likely to be seen by the user after it is recommended i.e. the recommendation
is not ignored by the user.

Train and Test: This data will be trained using 80% of training set
and tested using 20% of the training set.

 

 

 

 

x

Hi!
I'm Joan!

Would you like to get a custom essay? How about receiving a customized one?

Check it out